Population size is a key parameter for the conservation of animal species. Close-kin mark-recapture (CKMR) relies on the observed frequency and type of kinship among individuals sampled from the population to estimate population size. This approach requires being able to determine the age of sampled individuals. One common approach, particularly in fish studies, is to measure animal length and using an assumed age-length relationship (a ‘growth curve’). We used simulation to test the effect of misspecifying the length measurement error and the growth curve on population size estimation. Simulated populations represented two fictional shark species, one with a relatively simple life history and the other with a more complex life history based on the grey reef shark (Carcharhinus amblyrhynchos). We estimated sex-specific adult abundance, which we assumed to be constant in time. We observed small median biases in these estimates ranging from 1.35 to 2.79% when specifying the correct measurement error and growth curve, with true coefficients of variation between 21.56 and 28.50%. Introducing error via misspecified growth curves resulted in changes in the magnitude of the estimated total population, with upwards shifts negatively biasing abundance estimates. Over- and underestimating the length measurement error did not introduce a bias and had negligible effect on the variance in the estimates. Our findings show that assuming an incorrect length measurement error has little effect on estimation, but having an accurate growth curve is crucial for CKMR whenever ageing is based on length measurements. If ageing could be biased, researchers should be cautious when interpreting CKMR results and consider the potential biases arising from inaccurate age inference.