Nowadays, users need to keep information in high-level of security by authentication tools. One of these tools is the keystroke biometric that we have studied for this master thesis. This authentication tool is based on habitual typing rhythm patterns of each person. The technology measures the time of pressing and releasing a key while the user is typing the password, then the system calculates four keystroke features (Up-Up time, Up-Down time, Down-Up time and Down-Down time). We have clarified how the system works and explains theoretical concepts of the approach. The approach is described using three types of methods (five-characters password with five samples, ten-characters password with five samples, ten-characters password with ten samples). In the implementation, the system collects and analyzes the keystroke features. Experimental results include tests with legitimate users and impostors who attack the system. The results were evaluated with realistic data and a combination of four features. The best results belonged to the second method by ten-characters password with five samples. It achieved an equal error rate (EER) of 1%.