To evaluate predictive factors for retrograde ureteral stent failure in patients with non-urological malignant ureteral obstruction. Between 2005 and 2014, medical records of 284 malignant ureteral obstruction patients with 712 retrograde ureteral stent trials including 63 (22.2%) having bilateral malignant ureteral obstruction were retrospectively reviewed. Retrograde ureteral stent failure was defined as the inability to place ureteral stents by cystoscopy, recurrent stent obstruction within one month, or non-relief of azotemia within one week from the prior retrograde ureteral stent. The clinicopathological parameters and first retrograde pyelographic findings were analyzed to investigate the predictive factors for retrograde ureteral stent failure and conversion to percutaneous nephrostomy in multivariate analysis with a statistical significance of p < 0.05. Retrograde ureteral stent failure was detected in 14.1% of patients. The mean number of retrograde ureteral stent placements and indwelling duration of the ureteral stents were 2.5 ± 2.6 times and 8.6 ± 4.0 months, respectively. Multivariate analyses identified several specific RGP findings as significant predictive factors for retrograde ureteral stent failure (p < 0.05). The significant retrograde pyelographic findings included grade 4 hydronephrosis (hazard ratio 4.10, 95% confidence interval 1.39-12.09), irreversible ureteral kinking (hazard ratio 2.72, confidence interval 1.03-7.18), presence of bladder invasion (hazard ratio 4.78, confidence interval 1.81-12.63), and multiple lesions of ureteral stricture (hazard ratio 3.46, confidence interval 1.35-8.83) (p < 0.05). Retrograde pyelography might prevent unnecessary and ineffective retrograde ureteral stent trials in patients with advanced non-urological malignant ureteral obstruction.