We aimed to predict residual tumor (R) classification and overall survival (OS) in patients who underwent first-line surgery for pancreatic ductal adenocarcinoma (PDA) using preoperative magnetic resonance imaging (MRI). 210 patients with PDA who underwent MRI and first-line surgery were randomly categorized into a test group (n=150) and a validation group (n=60). The R classification was divided into R0 (no residual tumor) and R1/R2 (microscopic/macroscopic residual tumor). Preoperative MRI findings associated with R classification and OS were assessed by using logistic regression and Cox proportional hazard models. In addition, the prediction models for the R classification and OS were validated using calibration plots and C statistics. On preoperative MRI, portal vein encasement (odds ratio 4.755) was an independent predictor for R1/R2 resection (P=0.040). Tumor size measured on MRI (hazard ratio [HR] per centimeter 1.539) was a predictor of OS, along with pathologic N1 and N2 stage (HR 1.944 and 3.243, respectively), R1/R2 resection (HR 3.273), and adjuvant chemoradiation therapy (HR 0.250) (Ps<0.050). These results have been validated. In conclusion, preoperative MRI was valuable for predicting R1/R2 resection using portal vein encasement, and was useful for predicting OS after surgery for PDA using tumor size.