Blood is essential to life also, the number of blood cells plays a significant role in observing an individual’s health status. Any disorder refers to various diseases, especially for white blood cells WBC which is the basic part of the immune system. Classification WBC subtypes is very useful for diagnosing disease, infection etc. Deep learning technologies can reduce the limita- tions in Medical settings and for the time it takes in classification tasks. Also, a lot of clinic experience is required for a doctor to detect the amount of WBC in human blood and classify it. The basic idea of the proposed study is to employ CNN models for classifying WBC into four subtypes (Eosinophil, Lymphocyte, Monocyte, Neutrophil). In this study, the transfer learning was employed for two techniques as follow: using VGG-16 and Mobile-net models for feature extraction and train the SVM and QDA classifiers with the extracted features. Also,we used VGG-16 and Mobile-net for feature extraction and classification using fine-tuning technique. For the first technique, the best performance in the classification of white blood cells is given by the hybrid model VGG-16 with SVM with the accuracy of 98.44%. For the second technique, the best result of 99.81% accuracy obtained from VGG-16 fine-tuned model.The experimental results showed that the use of the convolutional neural network models contributed to improv- ing the classification success of white blood cell subtypes.