The real life has been converted from real to virtual world called social networks. We, you, our families and friends are part of this virtual world. Even if you do not have an account on social networks, you are still a part of the virtual world in a way or another. This virtual world has succeeded in connecting the world all together crossing all borders between countries, societies, cultures and races. In real world we have our own rules, policies, privacy, and security in our home and society, and we have control over them. However, in the virtual world you have no control over rules and policies, but you are to abide them; your privacy and security are not well reserved. Trillions of data are transmitted every day on online social networks; no one can ensure that the transmitted data are secured. Virtual world have billions of members, millions of them are fake, some of them are using real person information and others using imaginary none existing personal information in the real world. The aim of this research is finding the best available solution for the detection of fake accounts problem on online social networks. In this research, a survey on sixteen researches studies that have proposed solutions for solving the detection of fake accounts problem have been done. Social network platforms, evalu ation metrics, machine learning algorithms and models, data scale, features and result accuracy were parts of our survey. We have devel oped our own outcomes metrics to evaluate the available solutions