OBJECTIVE To compare the estimated effect on birth weight of reductions in maternal cigarette consumption and urinary cotinine during pregnancy. METHODS An observational study of 641 women with complete data on cigarette consumption, urinary cotinine and infant birth weight. Correlation and regression analyses were used to examine relationships between birth weight, cigarette consumption and urinary cotinine at first and last prenatal visits. RESULTS Correlations of cigarette consumption and urinary cotinine with infant birth weight were -.23 and -.30 (first visit) and -.26 and -.31 (last visit); all P values were < .001. The regression equation relating urinary cotinine concentrations at first and last visits to infant birth weight explained a significantly larger proportion of the variability in birth weight than the equation relating cigarette consumption at these visits to infant birth weight, 11% vs. 7%, P = .04. Among continuing smokers, both equations predicted gains in birth weight in association with reductions in cigarette consumption, but quitting smoking before the first visit was associated with the most weight gain. As compared to the average infant birth weight of a woman who smoked 20 cigarettes per day throughout pregnancy, the estimated gain in birth weight would be 105 g if she cut down by 10 cigarettes per day after the first visit, 210 g if she quit after this visit and 310 g if she quit before the first visit. CONCLUSIONS For women still smoking at their first prenatal visit, infant birth weight is already compromised, but subsequent reductions in cigarette consumption are associated with gains in birth weight. For women who cannot quit smoking, these reductions need to be substantial if increases in birth weight of > 100 g are to be achieved.