The purpose for connecting the transistors in parallel is important because it helps to decide which parameters need to be preserved. First, are they BJT, JFET, PUT, IGBT or MOSFET? What is their cutoff frequency? Is the circuit operation linear or non linear (e.g. class A, AB, B, C, D, digital, etc.) Are the inputs or outputs tuned? Is the answer supposed to be specific for exactly two transistors, or the general case for n transistors? Is matching of transistors allowed or are we to assume a range of parameters? May we couple them thermally?
After these answers we might be able to discuss some details.