Phenotypic selection is commonly used in agroforestry, both in genetic improvement and as a component of "good practice" in seed collection. In the first case, the aim is to secure genetic gain. In the second case, selection is used to ensure that seed supplies meet given minimum quality standards, or that poor quality sources are avoided. Here we examine the effectiveness of phenotypic selection in natural forest stands of the Amazonian timber and multipurpose tree Calycophyllum spruceanum Benth.. We ask (a) whether mothertrees with high estimated annual height and diameter increments had faster growing progeny than mothertrees with low values; (b) whether forked mother-trees tended to have higher proportions of forked progeny than unforked trees; (c) whether spatially isolated mother- trees tend to produce slower growing progeny than mother-trees growing together with conspecifics. In each case, we found no evidence of differences between the respective groups. We offer explanations for these findings and discuss their implications for tree improvement and seed collection.