Pneumonia causes significant morbidity and mortality in Aboriginal populations in Australia's Northern Territory (NT). Kava, consumed in Arnhem Land since 1982, may be a risk factor for infectious disease including pneumonia. A case–control study (n=115 cases; n=415 controls) was conducted in 7001 Aboriginal people (4217 over 15 years). Odds ratios (OR) were calculated by conditional logistic regression with substance use and social factors as confounders. Pneumonia was not associated with kava use. Crude OR=1·26 (0·74–2·14, P=0·386), increased after controlling for confounders (OR=1·98, 0·63–6·23, P=0·237) but was not significant. Adjusted OR for pneumonia cases involving kava and alcohol users was 1·19 (0·39–3·62, P=0·756). In communities with longer kava-using histories, adjusted OR was 2·19 (0·67–7·14, P=0·187). There was no kava dose–response relationship. Crude ORs for associations between pneumonia and cannabis use (OR=2·27, 1·18–4·37, P=0·014) and alcohol use (OR=1·95, 1·07–3·53, P=0·026) were statistically significant and approached significance for petrol sniffing (OR=1·98, 0·99–3·95, P=0·056).