In this study, eight subjects were exposed in a simulated office to 31 combinations of indoor environmental conditions, assigned by orthogonal design and uniform design. Conditions comprised variations of Predicted Mean Vote (PMV), illuminance, sound pressure and CO2 concentration (independent of a consistent ventilation rate) as indicators of thermal, lighting, acoustic and indoor air quality. Participant satisfaction with each of the four factors and with overall environmental conditions were measured with a questionnaire. Multiple interactions were detected with a partial correlation analysis and regression analysis. Results showed an adjusted effect of illuminance on perceived acoustic environment, a significant effect of the thermal environment on indoor air quality satisfaction, and a slight effect of sound pressure on indoor air quality satisfaction. Linear and geometric mean regression models were investigated for predicting overall satisfaction from the factor satisfaction scores. For the linear model, it was determined that multicollinearity among factor satisfaction levels may result in non-significant and biased estimated coefficients. The geometric mean regression model provides better prediction accuracy than the linear regression model with fewer coefficients, and accounts for the finding that the lowest satisfaction level with any environmental factor appears to drive overall satisfaction.