A comet flyby, like the one planned for the Comet Interceptor mission, takes place under conditions that remain largely unknown up to the moment of encounter. A detailed trajectory design phase, which includes verification of the technical limitations implied by the flyby geometry, precedes target comet selection. Thus the flyby velocity and the solar zenith angle at closest approach are known in advance. Solar irradiance and the neutral gas expansion speed can be estimated reasonably well. However, the comet outgassing rate, the dust production rate, and the solar wind conditions are only known within broader uncertainty margins. The present paper aims to optimally choose the one degree of freedom that is available for tuning the flyby conditions: the distance of closest approach. This choice is based on a simplified formalism that expresses, on one hand, the science return to be expected as a function of the closest approach distance, and, on the other hand, the risks implied by a close approach. This is done by performing Monte Carlo simulations over a large sample of possible comet flyby configurations, based on the expected probability distributions of the gas and dust production rates and the solar wind conditions, for different closest approach distances. For small flyby distances, a spacecraft has the opportunity to study the nucleus, the neutral gas coma, and the induced magnetosphere from up close, benefiting the science return. There is a trade-off to be made against the cometary dust collision risk, which becomes larger close to the nucleus. This trade-off is illustrated for the case of the Comet Interceptor main spacecraft and the two probes it plans to release. The change of the optimal flyby distance with gas and dust production rate, solar EUV flux, and flyby speed is discussed.