Using a sample of individuals who hold target retirement funds (TRFs), we examine how people use arithmetic to estimate their retirement age. We find a robust “zero” bias where investors have a strong preference for TRFs that end with zero compared to TRFs that end with five. The evidence is consistent that the bias is an outcome of people using imprecise arithmetic, specifically rounding up and down in the computational estimation required to estimate their retirement year. The zero bias manifests itself in people born in years ending between eight and two. Those born in zero- through two-ending years select TRFs that imply they intend to retire at 70, whereas those born in eight- and nine-ending years choose TRFs that imply retiring at 60. The choices can significantly lower or increase wealth by altering the contribution amounts and exposing investors to risk incompatible with their age profile. The bias is particularly costly for those who are risk averse and select later TRFs but is also most beneficial to risk-averse consumers who choose early TRFs. We experimentally confirm that the contribution rates are related to the TRF choices and that the use of imprecise mathematical rounding is implicated in the bias.