Sumatra island is the island with the highest theft rate in Indonesia. This study aims to determine the factors that have a significant effect on number of theft crime on the Sumatra Island from 2016 to 2021 by using panel data regression. This research uses panel data regression method which is a combination of cross section and time series data. The data used is secondary data obtained from Central Bureau of Statistics. Research variables include population density (X1), average years of schooling (X2), case completion rate (X3), and number of drug abuse (X4). The best model estimation was carried out for three models, including Common Effect Model (CEM), Fixed Effect Model (FEM), and Random Effect Model (REM). The selection of the best model of three models was carried out through the Chow test, Hausman test, and Lagrange Multiplier test. The best model result is individual effect FEM, with R2 = 94.8%. After partial test, the variables that have significant effect on number of theft crime in Sumatera Island are X2, X3, X4. The MAPE value of the best model is 23.15% which shows a feasible category.