Objective: To explore psychological health of nurses in terms of depression, anxiety and stress with predictive effect of demographic characteristics. Methods: It was a cross sectional study in which 300 on duty male and female nurses were included from government and private Hospitals of Gujrat, Pakistan. A demographic sheet and Urdu version of Depression, Anxiety & Stress Scale (DASS-21) were used to collect data. Descriptive statistics, two steps cluster analysis and neural network modeling were carried out by using SPSS 21. Result: Two steps cluster analysis classified the data in two clusters. Cluster 1 revealed low (f = 43) and medium (f = 61) level of depression, moderate (f = 81) and severe (f = 23) level of nervousness and mild (f = 20) level of stress, respectively. Whereas, in cluster 2 all the participants showed normal level of depression, anxiety and stress. Furthermore, neural network modeling revealed significant predictive importance of demographic variables. The most important predictor of mental health of nurses was attended COVID-19 patient 0.460 (100% normalized importance) next was age 0.154 (33.6% normalized importance), duty timing 0.029 (28.2% normalized importance), education .126 (27.5% normalized importance), marital status .101 (21.9% normalized importance), gender .020 (4.4% normalized importance) and nature of job .010 (2.1% normalized importance) respectively. Conclusion: Nurses during COVID-19 were facing mild and moderate level of depression and moderate and severe level of anxiety. Certain demographics features (attended COVID-19, age, duty timing & marital status) showed important predictive association with mental health of nurses. Managing the importance of predictive factors can be helpful to diminish the level of depression and anxiety in nurses.