One of evolutionary biology's most persistent puzzles is the fact that apparent directional selection on a heritable trait in a natural population often does not produce a selection response. We tested three possible explanations for this problem using data on body size of more than 23,000 individuals, measured over 18 yr, in a collared flycatcher (Ficedula albicollis) population. Using a restricted maximum likelihood "animal model," we found a narrow-sense heritability for fledgling tarsus length of [Formula: see text] SE and substantial common environment effects ([Formula: see text] SE). For survival to adulthood, the selection differential on tarsus length was [Formula: see text] SE. There was, however, no response to this selection over the study period. One explanation for the lack of response might have been that selection was associated with only the environmental (nonheritable) component of phenotype, but we found significant selection on breeding values (the heritable component). There was also no evidence of fluctuating selection pressures or of antagonistic selection between the sexes in selection pressures. Thus, in contrast to earlier investigations in this same population, none of the potential explanations for the absence of a selection response was supported; we discuss alternative hypotheses yet to be investigated.