Rail tracks are subjected to high contact stresses leading to wear and fatigue. With increased axle weight and faster trains, there is a need for routine inspection of the rails. In order to improve punctuality and reliability of track rail inspection, intelligent systems and robotics can be employed. In this research, supported through the FP7 INTERAIL project, a robotic system is being developed for the detection and characterisation of rolling contact fatigue (RCF) defects on in-service track rails using alternating current field measurement (ACFM) techniques. The autonomous inspection system consists of a mechanised trolley, an FS02N Kawasaki robot arm and a commercially available AMIGO ACFM system, supplied by TSC Inspection Systems. The ACFM probe is used to both detect (as the trolley moves at a controlled speed with the sensor in a fixed position) and characterise (whilst the trolley is stationary using the robot arm to survey the rail head) RCF cracks. Results show that the ACFM sensor can detect RCF defects even at significant (approx 4 mm) lift-off; however, if lift-off exceeds 4 mm (for example due to rail-head wear), the data acquired may not be easily exploited to characterise the defects. This paper presents the results obtained based on the work carried out to date.