This paper considers a localization problem in time difference of arrival (TDOA)-based radio frequency identification (RFID) systems. To estimate the position of a target tag, this paper suggests three localization algorithms that use benchmark tags. The benchmark tags are the same type as the target tag, but either the locations or distance of the benchmark tags are known. Two algorithms use the benchmarks for auxiliary information to improve the estimation accuracy of the other localization algorithms such as least squared estimator (LSE). The other one utilizes the benchmarks as essential tags to estimate the location. Numerical tests show that the localization accuracy can be improved by using benchmark tags especially when an algorithm using the LSE is applied to the localization problem. Furthermore, this paper shows that our benchmark algorithm is valuable when the measurement noise is large.