Event-triggered approaches to control and estimation have the sensor transmit processed information when a measure of information ‘novelty’ exceeds a threshold. Prior work has empirically demonstrated that event-triggered systems may have significantly longer average sampling intervals than comparably performing periodically triggered systems. There are, however, few results that analytically characterize the tradeoff that event-triggering introduces between average sampling period and system performance. This paper examines that tradeoff for a sub-optimal solution to the constrained state-estimation problem considered by Xu and Hespanha. The sub-optimal solution is comparable to that used by Cogill, and extends the earlier work to unstable systems. In particular, the paper derives a sub-optimal solution that guarantees the specified least average sampling period. The paper also derives upper and lower bounds on the event-triggered estimator performance. Simulation results are used to demonstrate the utility of these bounds.