Feedback control in our injury device allowed the impactor to be sensitive to the biomechanical characteristics of the spinal cord and produce mechanically predictable injuries. We tested the hypotheses that (i) extracellular calcium [( Ca2+]e) in the rat spinal cord recovers with a time course dependent on the magnitude of injury intensity, (ii) [Ca2+]e is initially depressed at the injury epicenter to the same degree independent of injury severity, and (iii) acute (less than 3.0 h) recovery of [Ca2+]e to normal values occurs in that group of animals that shows only transient neurologic deficits in the postinjury period. Three levels of injury (light, intermediate, and heavy) were produced by controlling spinal displacement during the injury process. After injury, [Ca2+]e at the injury site decreased to values less than 0.1 mM and then recovered during the next 3 h. Incomplete recoveries occurred in the intermediate- and heavy-injury groups (0.72 +/- 0.01 and 0.58 +/- 0.01 mM, respectively). [Ca2+]e activity in the lightly injured group recovered to normal values by 3 h. Specific injury protocols therefore resulted in reproducible responses in the cellular microenvironment. Behavioral recovery could be predicted from mechanical impact parameters. Animals in the light-injury group had transient neurologic deficits in some behavioral tests (open-field walking) with no alteration in others (inclined-plane analysis). Neurologic tests that required coordination between fore and hind limbs (grid walking) did not reveal significant deficiencies until 14 days postinjury. Those animals in the intermediate and heavy groups showed initial and continuing neurological effects in all behavioral measures. It is therefore probable that acute mechanical descriptors and hypocalcia transients are predictive of the ongoing and subsequent pathology of spinal cord injury.