One way parents shape the fitness prospects of their offspring is by providing a nursery. When parents divide a brood across several nurseries, they must assess not only the costs and benefits of multiple nurseries, but also the optimal proportion of the brood to deposit in each. Here, we explored the factors shaping these decisions in the strawberry poison frog (Oophaga pumilio), a species that typically, but not always, deposits its offspring singly in small phytotelmata. By monitoring occupancy and tadpole success in an artificial phytotelm array used by a free-living population, we tested for preferences for nursery size and height, asked whether multi-tadpole deposition was non-random with respect to these factors, and assessed the fitness consequences of these decisions. Parents were equally likely to use all types of artificial phytotelmata, but multi-tadpole depositions occurred almost exclusively in large nurseries. The probability that a tadpole would complete metamorphosis was unrelated to the physical characteristics of the rearing site, and individual tadpole survival was equivalent regardless of whether the nursery held one, two, or three tadpoles. Multi-tadpole nurseries were more likely, therefore, to have one occupant surviving to independence (metamorphosis). Although tadpoles are usually deposited alone, O. pumilio mothers may benefit from the insurance function played by additional tadpoles in a nursery. While most animals that care for their young provide a single nursery for their entire brood, a few separate and rear each brood member in an individual nursery. In such cases, parents must decide not only which nurseries to use, but also how many young they should deposit into each. Here, we explore the factors shaping these decisions in the strawberry poison frog (Oophaga pumilio), a species that typically, but not always, rears its young individually in small water-filled leaf axils. While O. pumiliodo not appear to choose nurseries based on size, parents were more likely to deposit multiple tadpoles into larger nurseries. Additionally, because the number of tadpoles in a nursery did not influence likelihood of metamorphic success for each individual, nurseries that contained multiple tadpoles were more likely to produce at least one offspring that completed metamorphosis. These results highlight the fitness effects of plasticity in reproductive strategies, and the possibility of divergent parent and offspring fitness optima.