Individual fitness is a central evolutionary concept, but the question of how it should be defined in empirical studies of natural selection remains contentious. Using founding cohorts from long-term population studies of two species of individually marked birds (collared flycatcher Ficedula albicollis and Ural owl Strix uralensis), we compared a rate-sensitive (lambdaind) and a rate-insensitive (lifetime reproductive success [LRS]) estimate of individual fitness with an estimate of long-term genetic fitness. The latter was calculated as the number of gene copies present in the population after more than two generations, as estimated by tracing genetic lineages and accounting for the fact that populations were not completely closed. When counting fledglings, rate-insensitive estimates of individual fitness correlated better than rate-sensitive estimates with estimated long-term genetic contribution. When counting recruits, both classes of estimates performed equally well. The results support the contention that simple, rate-insensitive measures of fitness, such as LRS, provide a valid and good estimate of fitness in evolutionary studies of natural populations.