Regeneration in the central nervous system is assumed to be largely disabled by glial activity. We applied in vivo two-photon laser-scanning microscopy to investigate acute and subacute reactions of microglia and their interaction with lesioned axonal fibre tracts in the dorsal column of transgenic mice with cell-type specific fluorescent protein (FP) expression . In crossbreedings of TgN(Thy1-EYFP) and TgH(CX3CR1-EGFP) mice neurons and microglia could be distinguished by selective expression of yellow or green FP expression, respectively. In fully anaesthetized mice (for details see below) a laminectomy was done and small lesions were performed with a fine needle or with infrared laser pulses of different intensity in the dorsal columns at spinal cord segment L4. Subsequently, cellular responses were either continuously recorded for up to 11 h after lesion or were repeatedly observed over 2-5 days. Between the long-term experiments the dorsal wound was closed and mouse behaviour was closely followed for signs of infection or pain. Anaesthesia/analgesia: (1) single acute experiments: initially pentobarbital 80 mg/kg i.p., followed by methohexital-Na 50 mg/kg, i.v. (2) repeated experiments: during the experiments pentobarbital initially 80 mg/kg i.p., 20 mg/kg i.p. added about every hour as required; during the periods between the experimental inspections these mice received buprenorphin 0.05 mg/kg i.p. every 8-12 hours. Within 1-2 min after the lesion the microglia started to extend processes with protrusions at their end to the direction of the lesion. Subsequently, also some microglial somata moved into the same direction. After a few minutes lesioned axonal fibres generated bulbous endings. These bulbs were partly engulfed by microglia and removed. Microglia activity continued for the next days. The process was almost symmetric to the lesion site in the early acute state but was mainly rostrally orientated i.e. to the distal part of the axons in the subacute state. While microglial responses within the first hours after injury were mainly restricted to process extension towards the lesion site, long-term recordings over several days revealed a significant accumulation of numerous microglial cells starting at about 10 to 15 h post lesion and continuing for at least 5 days. The results show that the microglial cells play a continuous role in spinal cord injury, starting already within a few min and continuing for several days. Microglial responses can be dissected in distinct phases for which the molecular control mechanisms still have to be identified.