Published data on benzene vapor concentrations in work simulation settings were used to examine the predictive ability of a near field/far field vapor dispersion model with an exponentially decreasing vapor emission rate. A given simulation involved two 15-min periods of applying a known volume of benzene-containing liquid to equipment on a worktable in a room with a measured air exchange rate. Replicate personal breathing zone (15-min time-weighted average, TWA) and room area (1-hr TWA) air samples were collected. In our modeling, the benzene vapor concentration in the near field zone (at the worktable) represented the personal breathing zone exposure level, and the benzene vapor concentration in the far field zone represented the room area concentration. Across 10 simulation combinations of two factors (the mass of benzene applied and the room air exchange rate), the mean of the personal breathing zone exposure levels ranged from 0.2 to 9.9 mg m(-3), and the mean of the room area concentrations ranged from 0.05 to 5.05 mg m(-3). Our model provided reasonably accurate estimates of the measured benzene vapor concentrations. Linear regression of the mean measured personal breathing zone exposure versus the predicted near field concentration yielded slope = 0.93 and r(2) = 0.94; the null hypothesis that the true slope equals one was not rejected (p-value = 0.39). Linear regression of the mean measured room area concentration versus the predicted far field concentration yielded slope = 0.90 and r(2) = 0.94; the null hypothesis that the true slope equals one was not rejected (p-value = 0.20). Other statistical tests showed no significant differences between measured and predicted values. In addition, most predicted concentrations fell within an approximate range of one-half to twofold the respective measured concentrations.