Some wish to mandate preregistration as a response to the replication crisis, while I and others caution that such mandates inadvertently cause harm and distract from more critical reforms. In this article, after briefly critiquing a recently published defense of preregistration mandates, I propose a three-part vision for cultivating a robust and cumulative psychological science. First, we must know how to produce robust rather than fragile findings. Key ingredients include sufficient sample sizes, valid measurement, and honesty/transparency. Second, we must know how to identify robust (and non-robust) findings. To this end, I reframe robustness checks broadly into four types: across analytic decisions, across measures, across samples, and across investigative teams. Third, we must be motivated to produce and care about robust science. This aim requires marshaling sociocultural forces to support, reward, and celebrate the production of robust findings, just as we once rewarded flashy but fragile findings. Critically, these sociocultural reinforcements must be tied as closely as possible to rigor and robustness themselves—rather than cosmetic indicators of rigor and robustness, as we have done in the past.