During their 4th and 8th mo. of pregnancy, 155 women were interviewed about their drinking habits during the first 4 mo. of pregnancy. A comparison of quantity-frequency-variability (Q-F-V) and absolute alcohol (A-A) scores of the original responses, responses of 78 of these women 1 wk later and responses of all 155 women 4 mo. later, permitted estimates of the amount of variation at 4 mo. that could be attributed to forgetting. The cost of retrospective measurement, i.e., the effect of forgetting, was higher when a precise (A-A) rather than a more approximate (Q-F-V) measurement was used.