The purpose of this study was to determine whether 99mTc-mercaptotriacetylglycine (MAG3) can substitute for 99mTc-glucoheptonate (GH) in the detection of pyelonephritis. METHODS One hundred thirty renal scintigraphies were evaluated retrospectively in 38 children (21% boys, 79% girls; age range 1 mo-21 yr; mean age 7.2 yr) referred for evaluation during an acute clinical urinary tract infection and for follow-up studies. Twelve topographical regions were designated on each kidney. Each area was graded for severity of decreased radionuclide localization: mild (Grade 1), moderate (Grade 2) or marked (Grade 3). Early posterior views of MAG3 studies were compared to delayed posterior GH images. In all patients, both studies were performed on the same day. RESULTS Eighty-two studies were performed during an acute clinical infection and 48 were performed as follow-up. Seventy-seven percent of the studies had focal cortical lesions. Of all the cortical lesions identified by GH, MAG3 detected 74% (match lesions). A comparable percentage of lesions was identified in each region by both studies. GH scintigraphy detected 261 lesions (63 Grade 1, 149 Grade 2 and 49 Grade 3), and MAG3 detected 201 lesions (37 Grade 1, 117 Grade 2 and 47 Grade 3). MAG3 was unable to recognize 60 lesions identified by GH studies in 11 patients (mismatch lesions). Of these, 41% (26 of 63) were Grade 1, 21% (32 of 149) were Grade 2 and 4% (2 of 49) were Grade 3. In three cases, MAG3 identified lesions not seen by GH (reverse mismatch); all had acute symptomatic infection. CONCLUSIONS These data document that MAG3 in the early phase of the study (1-2 min) can detect Grade 2 to Grade 3 cortical lesions in patients with pyelonephritis, but it is less effective in detecting Grade 1 lesions.