Neurolytic celiac plexus block is a recognised treatment for visceral abdominal pain due to malignancy. The need for a diagnostic celiac plexus block prior to neurolytic blockade is of questionable value, as it may not predict a positive response and may incorrectly predict a lack of response. Our objective is to evaluate the efficacy of diagnostic celiac plexus block. The records of 59 patients treated with celiac plexus block during 1994-2000 were retrospectively reviewed. Diagnostic block was performed on 32 patients prior to the decision for subsequent neurolytic block (Group 1). Another 27 patients were directly treated with a neurolytic celiac plexus block (Group 2). Response of Group 1 to diagnostic and neurolytic blocks was compared. Data from Group 2 was used to project the response of Group 1 should those patients with negative response to diagnostic block proceeded to neurolytic block. A two-by-two table was then constructed. The diagnostic celiac plexus block predicted a positive response with a sensitivity of 93% and a specificity of 37%. The positive predictive value was 85% and the negative predictive value was 58%. The estimated "number needed to test" before a "true" nonrespondent to lytic block to be detected was 16.7. Therefore, a positive response to diagnostic block correlates positively with neurolytic celiac plexus block for abdominal visceral pain due to malignancy. However, diagnostic block is a poor predictor when the response is negative. Hence, its clinical role is questionable and may not be warranted for patients with terminal malignancy.