OBJECTIVE: To evaluate the diagnostic value of macroscopic haematuria for the diagnosis of urological cancers in primary care as well as referred patients. METHOD: Systematic review of published reports, identified by a search on Medline and FAMLI and by screening of reference lists of selected papers. The evaluation of the sensitivity was based on patients with proven cancer of the kidney, ureter, bladder, urethra or prostate. The evaluation of the positive predictive value (PPV) was based on patients complaining to their physicians of macroscopic haematuria. RESULTS: No study executed in a primary care setting was included. In referred patients the pooled sensitivity of macroscopic haematuria for bladder cancer, based on seven remarkably homogeneous studies, was 0.83 (95% CI = 0.80-0.85). For ureteral cancer, this was 0.66 (95% CI = 0.53-0.77) based on four reports and for renal cancer 0.48 (95%) CI = 0.36-0.60) based on three studies. The pooled PPV of haematuria for urological cancer was 0.22 (0.17-0.27) in referred patients. Most malignancies detected were bladder cancers (255/317). PPV was highest in patients aged 40 or more at 0.41 (95% CI = 0.10-0.78). CONCLUSIONS: The advice that all patients with macroscopic haematuria should receive a thorough diagnostic programme seems justified in a specialized setting, dealing with referred patients. At this moment no data are available to support or discourage a similar policy for GPs. Prospective studies on the diagnostic value of macroscopic haematuria for urological cancer in a primary care setting are urgently needed.