Population viability analysis (PVA) is recommended as a defensible framework for species at risk and habitat recovery planning. Data requirements of spatial PVAs, however, are intensive and parameter estimates and model structure are often uncertain. Sensitivity analyses of spatial and non-spatial input parameters can inform data needs and research priorities to help reduce uncertainty. We analyzed results of sensitivity analyses used in 95 spatial PVA models of plants and animals to identify influential parameters to prioritize costly research for species and habitat recovery planning. We focused on stochastic demographic PVAs of metapopulations using generic software published from 2000 to 2006. Out of 28 variables, the 9 most influential parameters on model predictions included the number of populations (88% of studies), metapopulation configuration (86%), vital rates (85%), catastrophes (84%), carrying capacity (80%), initial abundance (73%), dispersal survival (63%), dispersal rate (51%), and dispersal function parameters (50%). Although spatial parameters were the most influential, on average, only 13% of all studies varied these inputs. Relative to non-spatial parameters, spatial parameters were ranked highest in terms of data uncertainty and we recommend investing in research to increase their accuracy. Few spatial PVAs carried out comprehensive sensitivity analyses indicating a need for tools to measure spatial and non-spatial parameter influence on model predictions.