Sibling-size variation (SSV), estimated as the coefficient of variation of egg size, was investigated for 13 populations of brown trout Salmo trutta. SSV was negatively correlated with mean egg size both at the population and individual levels. After correction for the effect of mean egg size, SSV was also negatively correlated with stream size. These results provide new information about how salmonid SSV can vary at different ecological scales (individual, population and region). The results are discussed in light of competing theories for explaining SSV: (1) the passive effect hypothesis, stating that egg size variation follows passively from selection on egg size and (2) the bet-hedging hypothesis, stating that high SSV is adaptive in unpredictable environments.