Objective To examine the long-term retreatment rates and explore prognostic factors after percutaneous nephrolithotomy (PCNL) in the computed tomographic era. Methods Consecutive patients who underwent PCNL with a standardized technique attempting complete fragment detection and clearance by a single surgeon between September 2000 and June 2004 were identified. Through medical record, review details of procedures and outcomes were collected. Results A total of 166 renal units in 150 subjects were evaluated. Postoperative computed tomographic scans were conducted in 129 subjects. Median follow-up was 5.4 years (interquartile range, 4.2-6.2 years). Future ipsilateral procedures were performed in 23 renal units (14%) at a median of 2.9 years (interquartile range, 1.7-3.7 years). The cumulative retreatment rate at 7 years for noncalcium stones (24%) was not significantly higher than for calcium-based stones (14%; P =.07). Stone-free renal units had a lower cumulative retreatment rate (4%) than those with residual fragments <2 mm (33%) or ≥2 mm (30%; P =.001). When controlling for residual fragment size, renal units that were composed of uric acid or struvite had an expected hazard rate of retreatment 3.34 times larger than renal units composed of calcium oxalate/phosphate (P =.02) and renal units that were not stone free had an expected hazard rate 7.87 times larger than renal units that were stone free (P =.001). Conclusion In this population of complex stone patients treated by PCNL, there appears to be no such thing as an "insignificant fragment." Efforts at initial stone clearance should be maximized.