A novel car speed estimation technique from a pair of omnidirectional microphones is presented. This approach is based on the maximum likelihood principle and it directly estimates the speed without any assumptions on the acoustic signal emitted by the vehicle. This has the advantages of bypassing troublesome intermediate delay estimation steps as well as eliminating the need for an accurate yet general enough acoustic model. An analysis of the estimate for the particular case of a narrowband source is provided. The estimation algorithm is well suited to DSP implementation and performs well with preliminary field data.