A pressing issue in the emerging field of psychological network analysis is how to estimate a network on a sample selected by sum-score criterion. In psychological studies, it is common practice to select a sample based on the sum-score of the modeled variables (e.g., selecting based on symptom severity when investigating the associations between those same symptoms). However, this practice may introduce selection bias, distorting the association between variables in the resulting network. Here, we propose a new correction for selection bias in the Ising model, which involves estimating associations between variables while accounting bias due to the sum-score selection. In a simulation study, we demonstrate that our correction recovers the true network structure using both nodewise regression and multivariate estimation after a sum-score selection. We then show in a tutorial how to use our correction with four commonly used packages to estimate the Ising model, in which we have implemented our correction, namely IsingFit, IsingSampler, psychonetrics, and bootnet. For our tutorial we use empirical data on major depression symptoms from the National Comorbidity Study-Replication (N = 9,282), to also demonstrate the results of our correction with real data. After a sum-score selection (N = 2022), regular estimation resulted in an unusually sparse network, suggesting that associations between symptoms were biased, whereas this was not the case when we applied our correction. Future research could focus on correcting selection bias when estimating other popular network models, such as the Gaussian graphical model, to further ensure the validity of findings obtained after a sum-score selection.