Summary • Experimental field data and interspecific comparative analyses were used to detect effects of seed mass on seedling performance within and among seven species of Psychotria sown in gaps and shaded rainforest sites. In addition we compared the effects of seed mass within and among species to detect concordance between the two ecological scales. We used two comparative methods: phylogenetically independent contrasts and cross-species correlations. • Among species, we detected weak evidence of a positive correlation between seed mass and the probability of emergence in the shaded forest, and no effects of seed mass in gaps. • Among species, no significant correlations between seed mass and either seedling survival or seedling recruitment were found in any habitat. Other variables specific to each subgenus appear to be more important than seed mass in determining survival in the shaded forest. • There was a negative correlation between seed mass and relative growth rate (RGR) in both habitats. In gaps, small-seeded taxa exhibited particularly high RGR, compensating for the initial advantages of higher seed mass. • All species studied exhibited recruitment in gaps equal to or higher than that in the shaded forest. However, recruitment success in shaded forest relative to gaps increased with seed mass, indicating a higher affinity for shaded forest among larger-seeded taxa, but this relationship was only detected using PICS analysis. •6. Correlations between seed mass and seedling mass are similar within and among species, indicating a simple principle of mass transference. In contrast, correlations between seed mass and seedling emergence, seedling survival, seedling recruitment and RGR depend on the scale at which they are observed. Functional Ecology (2005) 19, 707–718 doi: 10.1111/j.1365-2435.2005.00984.x