The occurrence of positive cloud-to-ground lightning discharges in thunderstorms in the vicinity of Kathmandu, Nepal (27.4 ¿N, 85.2 ¿E) has been recorded during the 21-month period from March 1987 to November 1988 inclusive, using the daily registrations of a CGR3 lightning flash counter that provided records of the approximate numbers of positive ground flashes, negative ground flashes, and cloud flashes, with effective ranges of about 12 km for cloud flashes, 14 km for negative ground flashes, and 16 km for positive ground flashes. The detection of positive ground flashes was based on the presence of a relatively large negative-going step in the electric field change caused by the positive return stroke in the discharge. It was found that the overall mean proportion of positive ground flashes to all ground flashes was 0.28, and that the overall mean positive ground flash density was 0.6 km-2 yr-1. The uncertainty in these values is about ¿50%. The proportions of positive ground flashes in three of the climatic periods in the year were 0.26 in the premonsoon period (February to May inclusive), 0.34 in the monsoon period (June to August inclusive), and 0.38 in the postmonsoon period (September to November inclusive). This apparent increase in the proportion is accompanied by a seasonal decline in electrical activity, the premonsoon period being the most active, and the postmonsoon the least active. The increase also appears to be associated with a decrease in the proportion of cloud flashes to total flashes. The proportion of positive ground flashes is larger than is observed in most low-altitude tropical and subtropical sites but is smaller than that observed in winter thunderstorms at some higher-latitude sites. Possible reasons for the relatively high rate of occurrence of positive ground flashes (unsupported by concurrent atmospheric soundings) are the site altitude, thundercloud charge heights, vertical wind shear and the mountainous nature of the terrain. ¿ American Geophysical Union 1993