The aim of this study was to develop and evaluate a method for the quantification of inhomogeneity in ventilation scintigraphy. Ten healthy volunteers and 10 emphysematous patients were investigated. Anteroposterior (AP) and Posteroanterior (PA) images of planar ventilation scintigraphy were acquired. Lung regions of interest (ROI) were obtained by manual delineation of the lung contours and then divided into several 10-pixel-high horizontal-stripe regions. By allowing for the statistical noise of the pixel count rate, the biological coefficient of variation (CVB) of the pixel counts in each stripe region was calculated. The apex-to-base distribution of the CVB in the emphysematous lungs dispersed largely and with higher values than the corresponding distribution in the healthy lungs. The mean values of the CVB (MCVB), the ranges of the CVB (RCVB) and the maximum values of the CVB (MAXCVB) in the stripe regions in emphysematous lungs were significantly higher than the corresponding ones in the healthy lungs (all P < 0.001). The intraobserver variations of the MCVB, RCVB and MAXCVB (calculated using the standard deviations of the differences) were less than 2.3% units, 5.2% units and 3.9% units respectively. The corresponding values for interobserver variation were 5.7% units, 6.1% units and 6.4% units. A systematic decrease in lung ROI size, i.e. inclusion of successively less of the lung edge, resulted in a linear decrease of 1.7% units in the MCVB and MAXCVB of both emphysematous and normal lungs. In conclusion, the stripe-region method is a reliable tool for the quantification of inhomogeneity in the planar ventilation scintigraphy.