On the basis of data from 3 days in 1978 this paper investigates the relation between thunderstorm dimensions and electrical activity to determine the relative importance of thunderstorm size and thunderstorm environment on the sferics rate. Sferics were recorded continuously by a wide-band (100 ¿ 50 kHz) crossed-loop radio direction finder located at the radar site; precipitation data during the corresponding period came from the radar volume scan. The number of sferics associated with specific storms and recorded in 5-min intervals varied from a few tens to a few thousands, depending on the dimensions of the source storm and on its distance from the sferics receiver. For four thunderstorm situations this variation was reduced to a small scatter, with standard deviations corresponding to factors of 1.17-1.47, by fitting the data with the empirical relation: S ∝ Ax ¿ 10 yH ¿ 10-zr, where S is the observed number of sferics per 5 min, A the cross-sectional area, H the height, and r the range of the storm. For three of the four sets of data analyzed, x was just over unity, giving a nearly linear relation between the sferics rate and the storm's area; for a supercell storm the relation was higher than linear. For thunderstorms between 50 and 300 km distant the number of recorded sferics decreased exponentially, by a factor of about 2 every 45 km, or with z = 0.0068 ¿ 0.0011 per km. Thunderstorm height appeared the dominant parameter determining the rate of electrical activity; values of y for the 3 days were between 0.13 and 0.28 per km, the greater effect of thunderstorm height was found associated with a stronger wind shear in the 7.5- to 12-km layer.