Previous research indicates that physical examination cannot diagnose sacroiliac joint (SIJ) pathology. Earlier studies have not reported sensitivities and specificities of composites of provocation tests known to have acceptable inter-examiner reliability. This study examined the diagnostic power of pain provocation SIJ tests singly and in various combinations, in relation to an accepted criterion standard. In a blinded criterion-related validity design, 48 patients were examined by physiotherapists using pain provocation SIJ tests and received an injection of local anaesthetic into the SIJ. The tests were evaluated singly and in various combinations (composites) for diagnostic power. All patients with a positive response to diagnostic injection reported pain with at least one SIJ test. Sensitivity and specificity for three or more of six positive SIJ tests were 94% and 78%, respectively. Receiver operator characteristic curves and areas under the curve were constructed for various composites. The greatest area under the curve for any two of the best four tests was 0.842. In conclusion, composites of provocation SIJ tests are of value in clinical diagnosis of symptomatic SIJ. Three or more out of six tests or any two of four selected tests have the best predictive power in relation to results of intra-articular anaesthetic block injections. When all six provocation tests do not provoke familiar pain, the SIJ can be ruled out as a source of current LBP.