The performance of integrating detectors with either destructive read-out or non-destructive direct read-out (N-DRO) capabilities are investigated for an environment subject to charged particle and radiation hits. The optimal integration time for maximizing performance in the presence of a given hit rate is determined. This optimization is a result of striking a balance between achieving a sufficiently long integration time to reduce the effects of the “read noise” of the device while remaining short enough to keep the probability of a hit relatively low. The performance relative to that expected for background limited performance (BLIP) is evaluated using a combination of analytic techniques and Monte Carlo simulations. N-DRO arrays are considered in the context in which over a fixed clock time: (a) all data between hits is good and kept; (b) only the longest integration time for a given detector is kept; (c) only the data before the first hit is kept; (d) the performance of the detectors degrades with each hit; (e) the detector requires some time after a hit to recover. Best performance is obtained with case (a) of the N-DRO array in which all data is used. For instance whenever the time between hits is comparable to the time to become background limited, case (a) is a factor of 1.7 better in performance than the destructive read-out array (a factor of 2.9 in time). When degradation due to hits (case d) or due to a finite recovery time after a hit (case e) occurs, the performance lies between that expected for cases (a) and (c). These results are applied to estimating the performance degradation of the long-wavelength spectrometer of the Space Infrared Telescope Facility due to cosmic rays. The performance may be a factor of 1–5 worse than BLIP depending on the read noise and how performance is degraded due to particle hits.