Introduction Hepatic sinusoidal obstruction syndrome (SOS) is a potentially fatal complication after hematopoietic stem cell transplantation (HSCT). Current literature and clinical guidelines support that changes in spectral Doppler occur late in the disease course and imaging is thus only useful for confirming the diagnosis of SOS. Objective Our primary objective was to examine abdominal ultrasound variables as potential predictors of development of SOS in pediatric HSCT patients. Our secondary objective was to examine these variables as potential predictors of death due to SOS. Methods A single-center cohort retrospective study was conducted on patients ages 0-21 years who underwent HSCT between September 2001 and May 2017 at our institution. Patients were excluded if they did not have abdominal ultrasounds following HSCT. Clinical, demographic, grayscale and spectral Doppler liver ultrasound findings were evaluated. SOS diagnosis was based on CIBMTR criteria and date of SOS diagnosis was the first date patient met the Seattle criteria (minus the 30 day limit). We modeled the odds of SOS diagnosis within 100 days after HSCT and SOS death as a function of each of the 15 ultrasound variables by fitting single-predictor logistic regression models and we fit a Cox regression model with ultrasound variables included as time-varying covariates. Results 333 patients received a HSCT. 140 subjects had ultrasound data available. 32 patients developed SOS and 9 of these patients died. The model estimates that the odds of SOS diagnosis more than double per one standard deviation increase in peak systolic velocity (PSV) in common hepatic artery (CHA) or left hepatic artery (LHA) and more than triple per one standard deviation decrease in main portal vein (MPV) velocity or change in ascites severity. Cutoff values could be chosen to yield sensitivity and specificity in the 70% range for all four variables. MPV velocity and ascites severity were the only two predictors for SOS death. Several ultrasound variables were statistically significant predictors in the Cox models for time to SOS diagnosis. Conclusion Several ultrasound variables can be used as predictors for a patient's risk of developing SOS. The strongest predictors are ascites severity, MPV velocity, CHA PSV and LHA PSV. Ascites severity and MPV velocity are also significant predictors of SOS death.