In this paper we study the determinants of NPL growth in banking system of Iran. Worsening of loan portfolio quality has been one of the biggest challenges of the economy in recent years. To identify the effects of different factors on NPL, at macro level and bank level, we estimate an unbalanced panel data model, LSDVC, Based on persistence of dependent variable and unbalanceness of our quarterly panel data of banks over 1381-1392. Regressors of our model are NPL, GDP growth, real lending rate and exchange rate, all of which are lagged for one period, as well as market share of private banks, loans to assets (LTA), size and loan growth rate of banks. While the authors find that exchange rate is insignificant, lagged NPL, market share of private banks, lagged GDP, and loan to assets has a significant positive effect on loan portfolio quality. In contrast, real lending rate, loan growth, and size of bank has a significant negative effect on NPL, thereby improve loan portfolio quality. However, square of market share of private banks and square of size have negative and positive effect, respectively.