This paper presents results from the first systematic survey of VLF wave activity at Halley, Antarctica (76 °S, 27 °W, L = 4.3). Beginning in 1971, the peak, average and minimum (P, A, M) signal levels observed in four frequency bands centred on 0.75 kHz, 1.25 kHz, 3.2 kHz and 9.6 kHz have been recorded every 5 min. At these frequencies the observed radio noise is largely natural, the waves being generated either in the magnetosphere (e.g. chorus, hiss, etc.) or near the ground, the latter principally from lightning discharges (radio atmospherics, or “spherics”) which reach the receiver after propagating some distance in the Earth-ionosphere waveguide (lightning does not occur in the immediate vicinity of Halley). Here we analyse the observations for 1984, the first complete year for which we have data in digital form, in terms of thunderstorm regions, as a benchmark for more extended studies of possible long-term change in global lightning activity. The data are presented in compressed colour graphic format which facilitates the identification of periodic (diurnal and annual) and aperiodic variations. At 3.2 kHz, attenuation in the Earth-ionosphere waveguide is severe, and only relatively few spherics, from close lighting source regions, are observed. Thus, whilst the 3.2 M channel is insensitive to lightning, and responds mostly to magnetospheric emissions, the 3.2 P channel is dominated by spherics. The 3.2 P data show a marked diurnal and seasonal variation symmetrical about Halley local noon and about the solstices, consistent with nearby sources and attenuation rates for subionospheric propagation which are much greater during the day than at night. At 9.6 kHz, waveguide attenuation is much lower (and there is less difference between day and night), and the minimum channel is dominated by a continuum of spheric noise originating from globally distributed distant source regions, notably those in the tropics. Consequently, there is no control by the local dawn-dusk terminator; the diurnal and seasonal variation is not symmetrical about Halley local noon and the solstices but consists of a quasi-sinusoidal diurnal variation, in which the phases of the minimum and maximum vary during the year: ~07 LT (LT∼-UT−2h at Halley) and ~17 LT in December (summer) and ~10 LT and ~21 LT in June (winter). Agreement between the observations and the CCIR (1983) empirical model is poor. A somewhat better fit is given by a simple model in which thunderstorm regions consist of point sources having radiated powers which vary with local time and season, the total effect at Halley being modelled as the sum of contributions from these sources.