Malaria is contagious and transmitted through Anopheles mosquito. This disease caused about 435,000 people died in 2017. Therefore, it is considered as a dangerous disease. This research aims to implement Siamese Convolutional Neural Network (SCNN) to identify malaria using cell images. SCNN is an architecture that uses two convolutional neural network with the same weight and parameters at each layer. Input data for this network are two paired images, where the first is the reference image and the second one is a test image and will produce a similarity score consisting of the numbers 0-1. The data uses consisted of two classes, such as parasitized and uninfected. Based on trials that have been carried out, we chose SCNN architecture consists of a pre-trained model VGG16 from Keras and a fully connected layer. Hyperparameter and parameter used are fixed and not all hyperparameter are used. Testing is done by changing the number of reference images and the number of training images. The best accuracy result obtained by using siamese convolutional neural network model is 94.35%.