Organisms must often make developmental decisions without complete information about future conditions. This uncertainty—for example, about the duration of conditions favorable for growth—can favor bet-hedging strategies. Here, we investigated the causes of life-cycle variation in Osmia iridis, a bee exhibiting a possible bet-hedging strategy with co-occurring one- and two-year life cycles. One-year bees reach adulthood quickly but die if they fail to complete pupation before winter; two-year bees adopt a low-risk, low-reward strategy of postponing pupation until the second summer. We reared larval bees in incubators in various experimental conditions and found that warmer—but not longer—summers, and early birth dates, increased the frequency of one-year life cycles. Using in situ temperature measurements and developmental trajectories of laboratory- and field-reared bees, we estimated degree-days required to reach adulthood in a single year. Local long-term (1950–2015) climate records reveal that this heat requirement is met in only ~7% of summers, suggesting that the observed distribution of life cycles is adaptive. Warming summers will likely decrease average generation times in these populations. Nevertheless, survival of bees attempting one-year life cycles—particularly those developing from late-laid eggs—will be <100%; consequently, we expect the life-cycle polymorphism to persist.