The concept of Darwinian fitness is central in evolutionary ecology, and its estimation has motivated the development of several approaches. However, measuring individual fitness remains challenging in empirical case studies in the wild. Measuring fitness requires a continuous monitoring of individuals from birth to death, which is very difficult to get in part because individuals may or may not be controlled at each reproductive event and recovered at death. Imperfect detection hampers keeping track of mortality and reproductive events over the whole lifetime of individuals. We propose a new statistical approach to estimate individual fitness while accounting for imperfect detection. Based on hidden process modelling of longitudinal data on marked animals, we show that standard metrics to quantify fitness, namely lifetime reproductive success, individual growth rate and lifetime individual contribution to population growth, can be extended to cope with imperfect detection inherent to most monitoring programs in the wild. We illustrate our approach using data collected on individual roe deer in an intensively monitored population.