Understanding the effects of coronal mass ejections (CMEs) requires knowing if and when they will impact and their properties upon impact. Of particular importance is the strength of a CME's southward magnetic field component (Bz). Kay et al. (2013, https://doi:10.1088/0004-637X/775/1/5, 2015, https://doi:10.1088/948 0004-637X/805/2/168) have shown that the simplified analytic model Forecasting a CME's Altered Trajectory (ForeCAT) can reproduce the deflection and rotation of CMEs. Kay, Gopalswamy, Reinard, and Opher (2017, https://doi.org/10.3847/1538-4357/835/2/117) introduced ForeCAT In situ Data Observer, which uses ForeCAT results to simulate magnetic field profiles. ForeCAT In situ Data Observer reproduces the in situ observations on roughly hourly time scales, suggesting that these models could be extremely useful for predictions of Bz. However, as with all models, both models are sensitive to their input parameters, which may not be precisely known for predictions. We explore this sensitivity using ensembles having small changes in the initial latitude, longitude, and orientation of the erupting CME. We explore the effects of different background magnetic field models and find that the changes in deflection and rotation resulting from the uncertainty in the initial parameters tend to exceed the changes from different magnetic backgrounds. The range in the in situ profiles tends to scale with the range in the deflection and rotation. We also consider a simple arrival time model using ForeCAT results and find an average absolute error of only 3 hr. We show that an uncertainty in the CME position of 8.1° ± 6.9° leads to variations of 6 hr in the arrival time. This measure depends strongly on the location of impact within the CME with the arrival time changing less for impacts near the nose.