Sera were obtained postmortem from 55 subjects classified into three groups; death due to asthma (FA, n = 21), asthmatic but death not due to asthma (NFA, n = 24) and a nonasthmatic control group (NAC, n = 10). A full autopsy was performed on all cases and a medical history, including details of allergies, was obtained by questionnaire from the next of kin. Grading of asthma severity by either questionnaire or autopsy was comparable (tP = 0.435, p > 0.05) and the mean pathology-grade was significantly higher for the FA group (3.375) compared to the NFA group (2.375), p < 0.05. Tryptase was elevated (> 2.0 micrograms/L) in 21/55 sera (38%) and there was no significant difference between the groups. ROC plots showed that tryptase levels did not discriminate between the FA and NFA groups, even if specimens were collected within 24 hours after death. Total IgE was significantly elevated in the FA group (geometric mean 140.3 kU/L) compared to the other two groups (NFA 30.2 kU/L, NAC 9.4 kU/L), p = 0.05. Fatal asthmatics also had a greater positivity (67%) to a screen for common inhalant allergens than did the other groups (NFA 30%, NAC 20%). Sera with a positive screen were tested against a panel of 10 common aero-allergens. Each sample was then assigned a number (N) and a score (S), dependent on either the number of allergens positive (N) or the total sum of pluses for all allergens (S). Both the N and S values were higher for the FA group (N = 98, S = 264) than the NFA group (N = 52, S = 151) and NAC group (N = 4, S = 8). The ratio (S/N) which gives an index (I) was 2.69, 2.90, and 2.00, respectively. Tryptase was poorly correlated to the total IgE level (r = 0.036); however, mean values for N and S were significantly different (N 6.81, S 4.50, and N 19.25, S 11.5, p < 0.05) for sera with tryptase levels < 2.0 or > or = 2.0 micrograms/L, respectively. We conclude that total and specific IgE may be useful predictors of asthma severity but that postmortem tryptase is not useful in the diagnosis of a fatal asthmatic attack.