Cognitive deficits are the most enduring and disabling sequelae of human traumatic brain injury (TBI), but quantifying the magnitude, duration, and pattern of cognitive deficits produced by different types of TBI has received little emphasis in preclinical animal models. The objective of the present study was to use a battery of behavioral tests to determine if different impact sites produce different patterns of behavioral deficits and to determine how long behavioral deficits can be detected after TBI. Prior to surgery, rats were trained to criteria on delayed nonmatching to position, radial arm maze, and rotarod tasks. Rats received sham surgery (controls), midline frontal contusions (frontal TBI, 2.25 m/sec impact), or unilateral sensorimotor cortex contusions (lateral TBI, 3.22 m/sec impact) at 12 months of age and were tested throughout the next 12 months. Cognitive deficits were more robust and more enduring than sensorimotor deficits for both lateral TBI and frontal TBI groups. Lateral TBI rats exhibited transient deficits in the forelimb placing and in the rotarod test of motor/ambulatory function, but cognitive deficits were apparent throughout the 12-month postsurgery period on tests of spatial learning and memory including: (1)reacquisition of a working memory version of the radial arm maze 6-7 months post-TBI, (2) performance in water maze probe trials 8 months post-TBI, and (3) repeated acquisition of the Morris water maze 8 and 11 months post-TBI. Frontal TBI rats exhibited a different pattern of deficits, with the most robust deficits in tests of attention/orientation such as: (1) the delayed nonmatching to position task (even with no delays) 1-11 weeks post-TBI, (2) the repeated acquisition version of the water maze--especially on the first "information" trial 8 months post-TBI, (3) a test of sensorimotor neglect or inattention 8.5 months post-TBI, and (4) a DRL20 test of timing and/or sustained attention 11 months after surgery. These results suggest that long-term behavioral deficits can be detected in rodent models of TBI, that cognitive deficits seem to be more robust than sensorimotor deficits, and that different TBI impact sites produce dissociable patterns of cognitive deficits in rats.