Background: Among preterm infants (< 12 months postnatal age and born < 37 weeks gestation), oscillometric systolic blood pressure (SBP) measurements are error-prone. However, some preterm infants have “gold standard” SBPs measured using an arterial line, in addition to the error-prone oscillometric SBPs. Objective: To compare different statistical methods to calibrate oscillometric SBPs to mimic arterial line SBPs Methods: We sampled participants from an existing pilot study of preterm infants treated with an antihypertensive therapy at the University of Alabama at Birmingham. Participants' data was obtained from the hospital's data warehouse for electronic medical records. Participants for this analysis had at least one arterial line SBP measurement, and for each of these "gold standard" measurements we identified the closest paired oscillometric SBP measurement (<10 minutes apart). We also collected participant factors such as race (white vs. black/African American), gender, gestational age at birth (weeks), postnatal age at arterial line SBP measurement (weeks), and whether the participant had received an antihypertensive medication. We predicted the arterial line SBPs from oscillometric SBPs and covariates using linear regression, linear regression with restricted cubic splines, random forest, and Bayesian additive regression trees (BART). Prediction errors were calculated using 10-fold cross-validation. We used complete case analysis. Results: We identified 822 measurements among 28 participants. Ten of 28 (36%) participants were black and 13 of 28 (46%) were female. The median (Q1, Q3) gestational age at birth was 33 (29, 35) weeks. The root mean squared error for predicting arterial line SBP from oscillometric SBP by itself was 17.1 mmHg. The root mean squared error of the predicted arterial line SPBs from different statistical methods varied little from the most error (linear regression with restricted cubic splines: 13.5 mmHg) to the least error (BART: 12.6 mmHg), meaning that any calibration resulted in at least a 21% reduction in error compared to the oscillometric SBPs. BART tended to underpredict the arterial line SBPs that were above 100 mmHg. Conclusions: Among preterm infants, statistical calibration is a viable strategy for reducing the variability of error-prone oscillometric SBPs. The different statistical models performed similarly in this small pilot study, but a larger sample size using additional important variables (e.g., birth weight) is required to confirm this finding.