A rapid method of urine screening and enterobacterial identification was evaluated. Results indicated that an average of 13.5 bacteria/oil immersion field (threshold value greater than or equal to 1) was observed in unsedimented urine of patients with significant bacteriuria, with an average of less than 1 bacterium/field in urines of patients without significant bacteriuria. In centrifuged urines, numbers of bacteria divided by amount of urine sedimented yielded similar results. Of 1758 urines studied, 136 yielded greater than or equal to 10(5) bacteria/ml, and 58 greater than 10(4) but less than 10(5) bacteria/ml, by conventional techniques. Gram-screening of unsedimented specimens gave sensitivity rates of 94.1%, specificity of 97.7%, and predictive positive and negative values of 78.5%, 99.5%, respectively; similar values were obtained with sedimented urines. Sensitivity rates of both screening methods for the 58 urines with greater than 10(4) but less than 10(5) bacteria/ml were 9.0%, 10.0%, respectively. Total correct enteric identification in 113 urines with positive screens and significant bacteriuria (greater than or equal to 10(5)/ml) was 82.3% and 90.3% with direct saline and broth Micro-ID methods, respectively. In 99 urines yielding pure or predominantly pure growth of 1 species of Enterobacteriaceae identification by direct saline and broth Micro-ID corresponded with isolated colony identification in 85.9%, 94.9% of cases, respectively. Gram-stain screening (together with back-up conventional plating in certain patient categories) and enterobacterial identification by direct broth Micro-ID, of urines with pure stains suggestive of greater than or equal to 10(5) Gram-negative rods/ml has been shown to be useful in laboratories without automated equipment for urine screening.