It really depends on exactly what you are comparing between the two simulators and simulation environments.

The best thing to do is to take the most basic simulation circuit that exhibits a distinct difference between the two simulators then start a process of troubleshooting to find the source of the discrepancy.

Discrepancies can be caused by
- model differences (are the device models and netlisted model parameters exactly the same?)
- simulator tolerance differences
- comparing results from different analysis types (i.e. comparing transient to harmonic balance)

To find differences you should start by comparing results using the same exact simulation type (DC, AC, transient) with the same tolerance settings. The simulation benches should be basic transistor parametric sweeps (DC IV and AC sweeps). If these simulations yield significant differences then you can dig into the model parameters to find out why. It could be the model parameters are different or perhaps there are subtle differences in how the models are implemented within the simulator code.

If the basic transistor simulations match then the issue is either you are comparing two different nonlinear simulators or the simulator tolerances are not set the same.

I can't provide a detailed troubleshooting procedure here, but the suggestions above are good place to start looking.