Among 1,226 pregnant women at Boston City Hospital, marijuana use during pregnancy was associated with a 79-gram decrease in birth weight and 0.5 cm shorter length after controlling for tobacco, alcohol, cocaine, and socioeconomic factors.
Pregnant women or those planning pregnancy who use cannabis, healthcare providers advising patients about prenatal substance use, and researchers studying developmental effects of cannabis exposure.
Marijuana-exposed infants weighed 79 grams less at birth — and the effect was stronger when urine testing detected use that women hadn't reported in interviews.
The Backstory
In March 1989, the New England Journal of Medicine published a study that would shape the conversation about cannabis and pregnancy for the next three decades. Barry Zuckerman, a developmental pediatrician at Boston City Hospital, had spent years watching pregnant women in his clinic navigate poverty, substance use, and inadequate prenatal care. He wanted to untangle which exposures actually mattered for their babies — and whether marijuana's effects could be separated from tobacco, alcohol, and cocaine.
What he found was modest in magnitude but powerful in implication: after controlling for every confounding variable he could measure, marijuana-exposed newborns weighed 79 grams less and were half a centimeter shorter. The numbers were small. The question they opened was not.
The Context: 1980s Boston
To understand why this study mattered, you need to understand what it wasn't. By the late 1980s, America was in the grip of the crack cocaine epidemic. Prenatal cocaine exposure dominated the research agenda and the public imagination — the "crack baby" panic was in full swing, with apocalyptic predictions about a lost generation of neurologically damaged children (predictions that would later prove vastly overblown).
Marijuana was an afterthought. Most pregnant women who used it also used other substances. Teasing apart marijuana's independent contribution to birth outcomes required a study designed specifically for that purpose — one that measured everything, controlled for everything, and used more than just asking women whether they'd used drugs.
Zuckerman understood a fundamental problem with prenatal substance research: women lie about drug use during pregnancy. Not maliciously — out of shame, fear of judgment, and in some jurisdictions, fear of criminal prosecution. Any study relying solely on self-report would systematically underestimate exposure and therefore underestimate effects. His solution was to combine structured interviews with urine immunoassays, creating a dual-detection system. The payoff was immediate: when analysis was limited to women whose drug use was only detected by urine testing (not interview), the associations with reduced birth weight were stronger. The women who wouldn't admit to using marijuana in an interview were, on average, exposed to more of it.
The Study Design
How They Did It
Prospective Cohort Design
From the general prenatal clinic at Boston City Hospital. Not a drug treatment population — a general obstetric clinic serving a low-income, predominantly minority urban population.
Structured interviews about substance use PLUS urine immunoassay testing, conducted both prenatally and postpartum. This caught women who denied use but tested positive.
Multiple regression models adjusted for tobacco smoking, alcohol use, cocaine use, maternal age, parity, race, education, income, prenatal care adequacy, and other covariates.
Birth weight, birth length, and head circumference recorded at delivery by clinical staff unaware of exposure status.
27% marijuana use and 18% cocaine use — much higher than previous interview-only studies suggested, validating the dual-detection approach.
Zuckerman et al. (1989), NEJM 320(12):762-768
The 27% marijuana prevalence was itself a finding. Previous studies relying on interviews alone had reported rates of 5-15% in similar populations. Zuckerman's dual-detection method roughly doubled the detected exposure rate — and demonstrated that the invisible portion of drug use was where the stronger effects hid.
The Numbers
Birth Outcomes
Marijuana-Exposed vs. Unexposed Infants
-79g
Birth weight
After controlling for tobacco, alcohol, cocaine, and socioeconomic factors
-0.5cm
Birth length
Shorter at delivery
-93g
Cocaine-exposed birth weight
For comparison — cocaine effect was modestly larger
-0.43cm
Cocaine-exposed head circumference
Marijuana showed no significant effect on head circumference
Zuckerman et al. (1989), NEJM
Let's be honest about what 79 grams means. The average newborn weighs about 3,400 grams (7.5 pounds). A 79-gram reduction is about 2.3% — roughly 2.8 ounces. For a healthy, full-term baby, this difference is almost certainly clinically insignificant. No pediatrician would worry about a baby weighing 7 pounds 4 ounces instead of 7 pounds 7 ounces.
But population-level effects don't work that way. The concern isn't the average case — it's the tail of the distribution. For infants already at risk of low birth weight due to prematurity, maternal malnutrition, preeclampsia, or other factors, an additional 79-gram reduction could push them below the 2,500-gram threshold that defines low birth weight and triggers additional medical concern. In Zuckerman's population — low-income, urban, with high rates of polydrug use and inadequate prenatal care — this marginal effect mattered more than it would in a low-risk suburban population.
What Came After: 35 Years of Confirmation
Research Timeline
Prenatal Cannabis Research After Zuckerman
Zuckerman et al. publish in NEJM
English et al. meta-analysis
Gunn et al. meta-analysis (BMJ Open)
National Academies of Sciences report
Corsi et al. (Nature Medicine)
Marchand et al. meta-analysis
Comprehensive meta-analysis (~7.9M women)
Multiple sources
The trajectory is clear: every major review and meta-analysis has confirmed Zuckerman's core finding, often with larger effect sizes. The 2016 Gunn meta-analysis found a mean birth weight reduction of 109 grams — 38% larger than Zuckerman's original estimate — and a pooled odds ratio of 1.77 for low birth weight (meaning exposed infants were 77% more likely to be classified as low birth weight).
The most recent meta-analysis, current to March 2024 and encompassing approximately 7.9 million women, went further: prenatal cannabis exposure was associated with increased risks of low birth weight, small for gestational age, preterm delivery, NICU admission, decreased head circumference, and perinatal mortality.
The Rising Urgency
This trend is the reason Zuckerman's 35-year-old study remains critically relevant. Cannabis use during pregnancy has more than tripled in the US between 2002 and 2020, driven by legalization, declining risk perception, and increasing normalization. In states with legal recreational cannabis, the rates are even higher. Some women use cannabis specifically for pregnancy-related nausea, viewing it as a "natural" alternative to pharmaceutical antiemetics.
The risk perception gap is the most concerning finding in recent surveys: many pregnant women who use cannabis believe it poses no risk to their baby. The legality of cannabis, combined with its "natural" image, creates a false sense of safety that Zuckerman's data — and the three decades of confirmatory research that followed — directly contradicts.
The Confounding Question
Cannabis directly harms fetal development
THC crosses the placenta freely. The endocannabinoid system is active in fetal brain development from the first trimester (neuronal migration, synapse formation, cortical organization). Disrupting this signaling at critical windows is biologically plausible as a cause of growth restriction. Animal studies confirm dose-dependent fetal growth reduction.
Confounding explains some or all of the effect
Women who use cannabis during pregnancy differ from non-users in dozens of ways: higher rates of tobacco and alcohol use, lower incomes, more mental health conditions, more stress, less prenatal care. Adjusting for known confounders reduces the effect, but unmeasured confounders may explain the rest. No randomized trial is possible.
Ongoing debate in perinatal epidemiology
This debate cannot be resolved by any observational study design, no matter how large or well-controlled. Randomizing pregnant women to use or not use cannabis is obviously unethical. Zuckerman did the best that was possible in 1989, and his successors have done the best that's possible since — but "association after controlling for known confounders" is permanently weaker evidence than a randomized trial would provide.
What tips the balance toward causal concern is the biological plausibility. The endocannabinoid system isn't just present in the developing brain — it's a primary guidance system for fetal neurodevelopment. CB1 receptors direct neuronal migration. Endocannabinoid signaling regulates synapse formation. Flooding this system with exogenous THC during critical developmental windows is mechanistically concerning in ways that go beyond statistical association.
Barry Zuckerman: The Bigger Story
Barry Zuckerman's career extended far beyond this single paper. As Professor and Chair of Pediatrics at Boston University School of Medicine, he co-founded Reach Out and Read (now the largest childhood literacy program in the US, operating in over 6,400 health centers), Medical-Legal Partnership (integrating legal services into healthcare), and Health Leads (connecting patients to essential resources). His 250+ publications consistently focused on the same question: how do maternal circumstances during pregnancy and early childhood shape lifelong health?
The 1989 NEJM paper was a piece of that larger project. Zuckerman wasn't an anti-drug crusader — he was a developmental pediatrician trying to identify which prenatal exposures mattered, how much, and what could be done about them. The study's measured tone and careful acknowledgment of limitations reflect that perspective.
What This Means Today
- Using cannabis for morning sickness high
Some women use cannabis to manage pregnancy-related nausea, not realizing it carries risk. Safer, well-studied alternatives exist (vitamin B6, doxylamine, ondansetron). Discuss with your provider.
- Occasional use in first trimester high
The endocannabinoid system is most active in early fetal brain development. Even occasional THC exposure during this window may disrupt neuronal migration and cortical organization.
- CBD-only products during pregnancy moderate
CBD has not been studied extensively in pregnancy. Many 'CBD-only' products contain trace THC. The precautionary principle applies. ACOG advises against all cannabis products during pregnancy.
- Secondhand cannabis smoke exposure low
Less studied, but THC can be absorbed through passive exposure. Pregnant women should avoid enclosed spaces where cannabis is being smoked.
ACOG Clinical Consensus No. 10 (2025); synthesis of evidence
Every major medical organization — ACOG, AAP, WHO — now recommends avoiding cannabis during pregnancy. This consensus rests on the cumulative weight of evidence that began with Zuckerman's 1989 study and has been reinforced by every subsequent meta-analysis. For women who are pregnant or planning to become pregnant, our guide to quitting during pregnancy covers the evidence and practical approaches to cessation.
Is 79 grams of birth weight reduction actually clinically significant?
For any individual healthy baby, probably not — it's about 2.8 ounces on an average 7.5-pound newborn. But at the population level, it shifts the entire distribution of birth weights downward, pushing more infants below the low-birth-weight threshold (2,500g). More recent meta-analyses suggest the true effect may be larger (109g in the Gunn 2016 meta-analysis). And for infants already at risk, every gram matters.
Does this prove marijuana causes lower birth weight?
No observational study can prove causation. What Zuckerman showed was an independent association after controlling for tobacco, alcohol, cocaine, and socioeconomic factors. The association has been replicated in dozens of subsequent studies and confirmed by multiple meta-analyses. Combined with strong biological plausibility (THC crosses the placenta and disrupts the fetal endocannabinoid system), the evidence is sufficient for medical organizations to recommend avoidance — even without proof of causation.
Has cannabis potency changed since this study?
Dramatically. Average THC content has increased from roughly 2-4% in the 1980s to 15-30%+ today, with concentrates reaching 80-90%. If the birth weight effect is dose-dependent (which animal studies suggest), modern cannabis could produce substantially larger effects than Zuckerman observed. This is one of the most urgent unanswered questions in prenatal cannabis research.
What about CBD during pregnancy?
CBD has not been adequately studied in pregnancy. Many CBD products contain trace THC. No medical organization endorses any cannabis-derived product during pregnancy. The ACOG Clinical Consensus (2025) explicitly recommends against all cannabis and cannabis-derived products during pregnancy and lactation.
Is cannabis use during pregnancy increasing?
Yes. Self-reported past-month cannabis use among pregnant women in the US tripled from 1.5% to 5.4% between 2002 and 2020. In states with legal recreational cannabis, rates are higher. Declining risk perception and increasing normalization are driving factors. Some women use cannabis specifically for pregnancy-related nausea, not realizing the potential risks.
What the researchers found
Marijuana use during pregnancy was independently associated with a 79-gram decrease in birth weight and 0.5 cm decrease in length after controlling for confounders including tobacco, alcohol, cocaine use, and socioeconomic status. The associations were stronger when drug use was detected by urine assay rather than self-report, suggesting that interview-only studies underestimate the true effect.
Why it matters
This NEJM publication established the scientific foundation for concerns about prenatal cannabis exposure. By using both interviews and biological testing to detect use (and showing that biological testing reveals stronger associations), it set a methodological standard and demonstrated that self-report alone is insufficient. Its findings have been broadly confirmed by subsequent meta-analyses showing cannabis-exposed infants have ~1.5-2.6x odds of low birth weight.
How the study worked
Prospective cohort study of 1,226 mother-infant pairs from the prenatal clinic at Boston City Hospital. Drug exposure was assessed through both structured interviews and urine immunoassays (conducted prenatally and postpartum). 27% tested positive or reported marijuana use; 18% for cocaine. Fetal growth outcomes (birth weight, length, head circumference) were assessed at delivery. Multiple regression controlled for tobacco, alcohol, cocaine, maternal age, parity, socioeconomic status, and other confounders.
Who was studied
N=1,226 mother-infant pairs recruited from a general prenatal clinic at Boston City Hospital. 27% reported marijuana use during pregnancy, 18% reported cocaine use.
What this study cannot tell us
Observational design cannot prove causation. Women who use marijuana during pregnancy differ from non-users in many unmeasured ways. Low THC potency of 1980s cannabis (typically 2-4%) limits generalizability to modern products. Self-report unreliable (the study itself showed this). No long-term developmental follow-up. Boston inner-city population may not generalize broadly. 79 grams is a modest effect (~2.3% of average birth weight) and may not be clinically significant in isolation. Confounding by polydrug use (27% marijuana, 18% cocaine overlap) is partially but not fully addressed.
How to read the evidence
Moderate evidence from a well-designed prospective cohort study published in the NEJM. The sample size (1,226) is adequate, and the use of biological verification strengthens exposure assessment. However, residual confounding is always a concern in observational studies of substance use during pregnancy. Subsequent meta-analyses have largely confirmed the direction of findings.
When this study was published
Published in 1989, reflecting cannabis potency and usage patterns from the mid-1980s. Modern cannabis is dramatically more potent (15-30%+ THC vs 2-4% in the 1980s), which could amplify effects on fetal growth. The fundamental biological mechanism (THC crossing the placenta and disrupting endocannabinoid signaling in fetal development) remains the same.
The bigger picture
Zuckerman 1989 launched a research trajectory that continues today. Subsequent meta-analyses (Gunn 2016, Marchand 2022, 2025 update) have confirmed the low-birth-weight association with pooled odds ratios of 1.5-2.6x. Meanwhile, cannabis use during pregnancy has tripled from 1.5% to 5.4% between 2002-2020 in the US, driven partly by legalization and declining risk perception. ACOG now explicitly recommends against cannabis use during pregnancy, citing this body of evidence.
Funding
NIDA/NIH grant R01-DA-03508
Conflicts of interest
None reported
Questions still open
- Does the effect size increase with modern higher-potency cannabis products?
- Is there a safe threshold of cannabis use during pregnancy?
- Does timing of exposure (trimester) affect the magnitude of growth restriction?
- Do the birth weight differences translate to meaningful long-term health differences?
- How does CBD-only use during pregnancy compare to THC-containing products?
Common questions
Is 79 grams of birth weight loss significant?
Does this prove marijuana causes lower birth weight?
Has cannabis potency increased since this study?
What do current medical guidelines say about cannabis during pregnancy?
Read the original research
Maternal marijuana and birth outcomes
New England Journal of Medicine, 320(12), 762-768
The New England Journal of Medicine is the most prestigious medical journal in the world, with the highest impact factor among general medical journals.
Citation
Zuckerman, Barry; Frank, Deborah A; Hingson, Ralph; Amaro, Hortensia; Levenson, Suzette M; Kayne, Howard; Parker, Steven; Vinci, Robert; Aboagye, Kwesi; Fried, Lisa E; et al.. (1989). Maternal marijuana and birth outcomes. New England Journal of Medicine, 320(12), 762-768. https://doi.org/10.1056/NEJM198903233201203