Abstarct: Snow is one of the important forms of precipitation at hydrological cycle of mountainous areas. Snow packs in mountainous basins is one of the most important and reliable water resources and hence, quantifying snowmelt has its valuable place in hydrology. Water from snowmelt can be stored as soil moisture, recharge groundwater and flow into rivers and lakes. Accurate prediction of snow water equivalent (SWE) quantities and timing is essential for water management. Deficiency in available snow measurements in mountainous basins due to lack of accessibility to few existing ground stations bold the necessity to utilization of indirect methods e.g. remotely sensed methods for snow-related studies. Remote sensing technology has many applications in various environmental studies including investigating snow and ice resources. Large area coverage of RS images, easy data acquisition at different scales and resolutions and fast processing of images using computers are among many advantages of remote sensing science for various hydrological studies. Of course using ground-baxsed data with satellite images on snow surveys can increase the efficiency of satellite data. The present study aimed to estimate the snow depth using satellite images and artificial neural networks (ANN) at Gachan snow station located in Ilam province. For this purpose, 111 daily snow cover MODIS product (MOD10A1) images from 2011- 2014 period were used. MOD10A1 images were acquired from the National Snow & Ice Data Center (www.nsidc.org) and after image processing, the input parameters of the artificial neural network were used as input. In this study, Multilxayer Perceptron (MLP) artificial neural network was used to model the relationship between derived variables from satellite images as its input and observed snow depth as its output. Comparing different artificial neural network architectures, it was deduced that the artificial neural network with best performance is a 14-neurons hidden laxyer. The correlation coefficient between the simulated and observed data was calculated as 0.98 and 0.99 for training and test data sets, respectively. With regard to acceptable performance of the artificial neural network, it can be used to monitor and predict the depth of snow at the study area, especially when the station is broke up and out of order. In these situations, the proposed method and trained ANN can be applied to estimate daily snow depth using MODIS snow product.