The inter-observer agreement (reliability) and validity of a clinical asthma severity scale (ASS) derived from separate scores of wheeze, heart rate and accessory muscle use (each on a 4-point scale) were studied in 60 children aged between 6 months and 17 years (mean 5.4 years). Independent assessments of these clinical parameters were made by two paediatricians, and they also rated patients as having a mild, moderate, severe or very severe acute episode (clinical judgement rating, CJR). Oxygen saturation (SaO2) was measured concurrently by a Biox 3700 pulse oximeter and readings were categorized as mild (SaO2 greater than or equal to 94%), moderate (91-93%) and severe (less than 91%). Agreement between clinicians was assessed by the weighted kappa statistic (kappa W). Agreement for the ASS score compared to the severity grade obtained from SaO2 was slight (kappa W = 0.34) and compared to CJR the kappa W was 0.55. An ASS score of moderate or worse (greater than 3) had sensitivity of 97% and specificity of 50% for prediction of admission. The maximum frequency and duration of nebulizer therapy following admission were significantly greater for severe patients than for moderate patients. Length of hospital stay did not reflect the ASS score in the emergency department but total duration of functional disability increased with ASS score. The substitution of an adjusted heart rate score for the raw heart rate score used in ASS detracted from scale performance. The ASS is an imprecise but reasonable quantitative measure of the severity of an acute episode of asthma.(ABSTRACT TRUNCATED AT 250 WORDS)