The planting of tree seedlings is a common restoration technique in the tropics, and using large-crowned, fast-growing shade species is recommended to suppress invasive grasses and accelerate forest succession. We analyzed the effectiveness of shade species in shading the forest floor during the rainy and dry seasons at young forest restoration sites, whether shade changes according to site for a given species, and whether crown architecture can predict the shade level. We measured the photosynthetically active radiation (PAR) intercepted by the tree crowns of 14 species in two 3-year-old restoration plantings. The ability to predict shade based on crown architecture traits was evaluated using multiple linear regressions. The interception of PAR varied according to species, site, and season for seven species and was generally higher during the rainy season. Low values of tree and first branch height and high values of trunk diameter and mean area of a leaf predicted greater light interception. For the dry season, the ability to predict PAR interception was weaker than that for the rainy season and affected by a shorter tree height and a greater crown area. The crown architecture of shade species did not completely predict their shading ability, and the preselection of shade species for forest restoration purposes based only on crown architecture traits is not effective. Therefore, it is important to consider other factors, such as how long trees retain their leaves throughout the year and the soil and management conditions of the sites undergoing restoration, during the selection of species.