In recent years it has become increasingly clear that most remedial technologies fail to completely remove dense nonaqueous phase liquid (DNAPL) from subsurface source zones. Recognition of this limitation leads to the question of what benefit can be achieved through partial removal of DNAPL. To address this issue, a mathematical technique referred to as the multiple analytical source superposition technique (MASST) has been developed. MASST is based on a conceptualization of a DNAPL source zone as a grouping of discrete subzones containing DNAPL (e.g., fingers and/or pools) separated by portions of the aquifer that are entirely free of DNAPL. Using analytical techniques, spatial superposition of responses to multiple sources is used to estimate aqueous mass transfer rates from individual subzones. This procedure accounts for multiple DNAPL subzones with different volumes, geometries, and locations within an overall source zone that is otherwise free of the nonaqueous liquid. The mass transfer rate from a particular subzone is affected by mass transfer from all other subzones in the vicinity. Groundwater flow is assumed to be uniform, and transport processes are considered to be at a steady state. Comparison of MASST results with exact analytical solutions and laboratory data confirms the validity of MASST. Sensitivity analyses indicate that source-zone architecture is a primary factor governing bulk mass transfer and source longevity. Analysis of rate-limited mass transfer within DNAPL subzones and advective-dispersive transport about DNAPL subzones indicates that advective-dispersive transport is the primary factor controlling mass transfer rates. Finally, results indicate that removal of the vast majority of the DNAPL will likely be necessary to achieve significant near-term improvements in groundwater quality. ¿ 2001 American Geophysical Union