Many species' range limits (RL) occur across continuous environmental gradients without obvious barriers imposing them. Such RL are expected to reflect niche limits (NL) and thus to occur where populations cease to be self-sustaining. Transplant experiments comparing fitness within and beyond species' ranges can test this hypothesis, but interpretive power depends strongly on experimental design. We first identify often overlooked aspects of transplant design that are critical to establishing the causes of RL, especially incorporating transplant sites at, and source populations from, the range edge. We then conduct a meta-analysis of published beyond-range transplant experiments (n = 11 tests). Most tests (75%) found that performance declined beyond the range, with the strongest declines detected when the measure of performance was lifetime fitness (83%), suggesting that RL commonly involve niche constraints (declining habitat quality). However, only 46% supported range limits occurring at NL; 26% (mostly geographic RL) fell short of NL with self-sustaining transplants beyond the range, and 23% (all elevational RL) exceeded NL with range-edge populations acting as demographic sinks. These data suggest an important but divergent role for dispersal, which may commonly constrain geographic distributions while extending elevational limits. Meta-analysis results also supported the importance of biotic interactions at RL, particularly the long-held assertion of their role in causing low-elevation and equatorial limits.