Materials have different densities so a number expressed in weight percent will be different than the same number expressed in volume percent. To compare the properties one of them has to be selected.

Another point is that the improvement of mechanical properties of a composite is related to the geometrical considerations (distribution of particles or fibres, their size,...) of the reinforcement in the matrix. That is why volumic percentage is closer to the expected effect this reinforcement may bring to the composite.

As said by Dr. Pawlowski, the parameters playing an important role like distribution, size, are mostly in relation with the volume %. 

The empirical expressions thay you will find in the literature for calculating the composite properties (mechanical : yield stress, strain, elastic modulus,... ; thermal : thermal conductivity,...) are mostly written by using the volume%.

Most theoretical formulae are derived for volume fractions of reinforcement, but weight fractions are easier to measure. As mentioned by the colleagues, the two can be calculated from each other using bulk density values, provided that the pore volume is zero.

Dr. Banhegyi's point is not to be underestimated - weight percentage may be easier to determine, but if you calculate the associated volume from mass and bulk density directly, you may introduce kind of a misconception if your reinforcement has porosity that is not filled by the matrix during the making of the composite (e.g. internal, closed pores). If that's a concern and of importance in your case, you can determine density e.g. via gas pycnometry to give you a value that is reduced in comparison to the theoretical one if non-accessible porosity is present.

Dear Dirk, I added exactly because of this that "provided that porosity is zero". Of course determining pore volume is a valuable additional datum.

Dear Gyorgy, yes, I did understand you that way, I merely wanted to stress your point a little more.

Be it wt.% or vol.%, the addition of reinforcement particles is going to influence the properties of the resultant MMC. It is easy to calculate and add reinforcements by wt.% especially if one uses in-situ techniques to produce MMCs. However, one needs to convert wt.% into vol.% to predict the properties as standard formulas have vol.% component.