From literature five methods for the detection of summer hail have been sel
ected. These five different methods have been tested on severe weather even
ts in the Netherlands that occurred during the summer of 1999. The general
trends in the scoring parameters of the detection methods as a function of
the warning threshold are rather similar, but there are substantial quantit
ative differences. Using a simple model, the effects of missing ground trut
h data on the scoring parameters of the detection methods has been describe
d qualitatively. It is concluded that, of all hail detection methods consid
ered, the method of Waldvogel performs best and is suited best for display
of the "probability of hail". (C) 2000 Elsevier Science Ltd. All rights res
erved.