The safety and efficacy of percutaneous biopsy of native kidneys performed entirely by nephrologists at the patient's bedside was evaluated in 101 consecutive patients. The location and depth of the kidney were determined with a portable ultrasound machine, and biopsy was performed with a 15G, automatic, spring-loaded biopsy device without direct ultrasonographic guidance. Renal tissue was obtained in 99 patients, and all samples were adequate for diagnosis, with an average of 33 glomeruli and more than 10 glomeruli in 97%. The number of biopsy attempts was four or fewer in 80% of patients. Three patients developed symptomatic bleeding, all of whom had a risk factor for bleeding, but none required procedures to control the bleeding. Asymptomatic hematuria occurred in two other patients. Overall, the mean decrease in hematocrit was 1.5, with a decrease of 5.0 or greater in six patients. The results are similar to those of previous studies using automatic devices but under direct ultrasound guidance. A subset of 20 patients with abnormal platelet counts, coagulation times, or bleeding times accounted for four of the five patients with complications. We conclude that percutaneous biopsy of native kidneys can be adequately and safely performed in its entirety by nephrologists at the patient's bedside. Furthermore, excellent results can be obtained without direct sonographic guidance. Hemorrhage occurs almost exclusively in those patients with abnormal platelet counts, coagulation times, or bleeding times.