A model to predict the onset of significant void (OSV) in vertical flow between parallel plates has been developed. The model was compared to the experimental data of Whittle and Forgan (1967) and Dougherty et al. (1990), showing excellent agreement. The model was also compared with the Saha-Zuber(1974) correlation, which has been widely used in computer codes for nuclear safety analysis. The present theory is more conservative than this correlation, and further shows that, contrary to this correlation, the Stanton number is not solely related to the Peclet number. This may explain the large error margins required for the Saha-Zuber correlation, and also the scatter beyond the error margins specified by the authors. The steady-state OSV heat fluxes for equal and unequal heating cases between parallel plates were compared. The arithmetic mean of heat fluxes for unequal heating cases is less than the heat flux for equal heating cases. The result may imply that OSV is controlled by local thermal parameters rather than bulk parameters.