It is well known that multi-group (MG) Monte Carlo (MC) calculations can run significantly faster than continuous energy (CE). Often high fidelity results are not required everywhere in the problem. This paper investigates an approach to accelerating MC calculation by combining MG and CE representations in a single MC calculation. An interface condition, based on slowing down calculation in an infinite lattice was developed to correct for resonance shielding while transitioning particles from regions with MG treatment to those with CE treatment. Moreover it was shown, that it is necessary to apply a different source normalisation in the CE and MG region to achieve reasonable accuracy of results. Prototype 1D implementation of the scheme was tested on small UOX and MOX problems. The error in fission rate distribution in CE region of combined MG-CE calculation was reduced to below 0.3% from up to 2% in a purely MG calculation. Prototype implementation was about 10% slower than theoretical maximum, obtained as a mixture of run-times of purely MG and CE problem, weighted by fraction of MG and CE tracking.