In an ongoing study of the hypothesised association between consumption of fish contaminated with persistent organochlorine compounds and low birth weight, a relevant measure of exposure is the accumulated fish consumption over long time periods. An issue of concern in this study is the quality of the retrospective exposure assessments. The present investigation was undertaken to determine the agreement between reported fish consumption in two interviews with approximately one year in between. in order to evaluate its impact on the effect estimates. The agreement decreased with increasing retrospective time interval between interview and focused time period of fish consumption. Furthermore, the exposure assessments based on the two interviews led to considerably different effect estimates.