Objective: To establish how different methods of estimating gestational age (GA) affect reliability of first-trimester screening for Down syndrome. Methods: Retrospective single-center study of 100 women with a viable singleton pregnancy, who had first-trimester screening. We calculated multiples of the median (MoM) for maternal-serum free beta human chorionic gonadotropin (free beta-hCG) and pregnancy associated plasma protein-A (PAPP-A), derived from either last menstrual period (LMP) or ultrasound-dating scans. Results: In women with a regular cycle, LMP-derived estimates of GA were two days longer (range -11 to 18), than crown-rump length (CRL)-derived estimates of GA whereas this discrepancy was more pronounced in women who reported to have an irregular cycle, i.e., six days (range -7 to 32). Except for PAPP-A in the regular-cycle group, all differences were significant. Consequently, risk estimates are affected by the mode of estimating GA. In fact, LMP-based estimates revealed ten "screen-positive" cases compared to five "screen-positive" cases where GA was derived from dating-scans. Conclusion: Provided fixed values for nuchal translucency are applied, dating-scans reduce the number of screen-positive findings on the basis of biochemical screening. We recommend implementation of guidelines for Down syndrome screening based on CRL-dependent rather than LMP-dependent parameters of GA.