In order to test the hypothesis that urine colour can be used as an index of hydration in critically ill patients, we selected 40 intensive care and high-dependency patients and correlated urine colour (scored on an eight-point scale) with various indices of hydration: urine:plasma sodium, osmolality and urea ratios, urine output and central venous pressure. In addition, we compared the colour-chart score with scores made by intensive care nurses (without the benefit of a colour chart) in order to test subjective assessment of urine colour. There were weak but statistically significant correlations between urine colour andurine output (Spearman's r = - 0.555) and between urine colour and urine:plasma sodium ratio (Spearman's r = - 0.459). Subjective assessment of urine colour appeared to be reliable. Thus, although urine colour does vary with hydration in thecritically ill, assessment of urine colour adds little to the overall assessment of hydration in this group of patients.