AIMS--To evaluate rapid organism identification on positive blood culture Bactec NR media (phial types 26, 27, 42 and 17), and to assess the usefulness of these procedures in a diagnostic microbiology laboratory. METHODS--Two hundred and sixty, first positive, blood culture bottles from individual patients were tested by rapid identification methods selected on the basis of Gram film organism morphology. Tube coagulase and latex agglutination were applied to presumptive staphylococci; latex agglutination antigen detection methods to suspected pneumococci, Neisseria and Haemophilus sp; and latex agglutination grouping tests for cultures thought to be non-pneumococcal streptococci. RESULTS--Media type did not influence test performance (p > 0.05 for all comparisons). Misapplication of methods occurred on eight occasions and there were 14 false positive results, nine involving the latex reagents for group C streptococci and pneumococci. The positive predictive values for tube coagulase tests and latex reactions for H influenzae type b, and N meningitidis groups B and C were 100%. The pneumococcal and staphylococcal latex tests gave positive predictive values of 94.1% and 62.5%, respectively, and the corresponding figure for streptococcal grouping reactions was 75.9%. With the exception of staphylococcal latex testing (80%) all investigation negative predictive values were > 90%. CONCLUSIONS--The performance of the staphylococcal latex agglutination method was unsatisfactory and it is not appropriate for use with the media studied. In view of the cross-reactions observed with the tests used to identify group C streptococci and pneumococci, positive findings must be interpreted with caution. In all other regards the protocol evaluated produced rapid, reliable, clinically useful information and, subject to local experience, is recommended to users of Bactec NR media.