We studied habitat selection by Piping Plovers (Charudrius melodus) at Lake Diefenbaker, Saskatchewan, during the 1992 and 1993 breeding seasons. Habitat selection was investigated at three levels: beach, territory, and nest site. Nest beaches had significantly more gravel substrate than non-nest beaches ( F I l , l l I = 21.52, p < 0.01). There was no significant difference in width between nest and non-nest beaches. All Piping Plover nests were located on gravel, though this substrate only accounted for approximately 43% of nest beach area. Nest sites were significantly farther from the nearest water (FI,, ,,,, = 12.24, p < 0.0 l), the main water body (Fl ,, ,,, = 6.55, p < 0.05), and had significantly more stones (FI,.,,,, = 10.98, p < 0.01) and rocks (Z = 2.93, p < 0.01) than random sites. Distance to the nearest water source, number of stones per square metre, and distance to the vegetation line were the three variables used by the stepwise discriminant function analysis to correctly classify 41.8% of nest sites and 87.1 % of random sites. This was still significantly better than chance (Z = 3.66, p < 0.01). The low percentage of correctly classified nest sites may indicate that some plovers at this lake used nest sites on lower quality substrate. Flooded nests were significantly closer to water than successful nests and depredated nests were closer to the vegetation line than successful nests.RCsumC : Nous avons etudie I'habitat du Pluvier siffleur (Churudrius melodus) au lac Diefenbaker, Saskatchewan, au cours des saisons de reproduction de 1992 et 1993. Le choix de I'habitat a ete etudie sous trois aspects, le choix d'une plage, le choix d'un territoire et le choix d'un site de nidification. Les plages choisies pour la nidification comportaient significativement plus de gravier que les plages sans nids (F,, ,, ,, = 2 1,52, p < 0,Ol). 11 n'y avait pas de difference significative de largeur entre les plages avec nids et les plages sans nids. Tous les nids etaient construits sur du gravier, meme si ce substrat ne constituait que 43% environ de la surface des plages avec nids. Les sites de nidification ktaient significativement plus eloignes de la masse d'eau la plus proche ( F l l , 1 8 1 1 = 12,24, p < 0,Ol) ou de la masse d'eau principale (F,,,,,,, = 6,55, p < 0,05) et comportaient significativement plus de cailloux (F,,,,6,1 = 10,98, p < 0,Ol) et de pierres (Z = 2,93, p < 0,Ol) que des sites choisis au hasard. Au cours d'une analyse par itapes des fonctions discriminantes, trois variables, la distance jusqu'a la masse d'eau la plus proche, le nombre de cailloux par metre carre et la distance jusqu'a la ligne de vegetation, ont permis de classifier correctement 41,8% des sites de nidification et 87,1% des sites aleatoires. I1 s'agit la d'un resultat plus satisfaisant que la determination au hasard (Z = 3,66, p < 0,Ol). Le faible pourcentage des sites de nidification classifies correctement peut signifier que certains pluviers de ce lac utilisent des sites a substrat de qualite moindre. Les nids inondks eta...