The tire noise emission is never omnidirectional as it is generally assumed when modeling using road traffic noise calculation models. Such assumptions are made due to poor availability of vehicle directivity data. This could contribute to noise level prediction inaccuracies. The results of performed noise emission directivity measurements (using five various passenger car tires rolled on eight various 'normal-noise' and 'low-noise' pavements with four test speeds) allowed to determine significant number of directivity patterns for various combinations of tire, pavement and speed. The following conclusions can be formulated: The influence of speed on tire noise emission directivity pattern is very dependent on tire/pavement combination. For non-porous pavements various correlations can be determined depending on tested tire. For porous and poroelastic pavements there are no correlation between speed and changes in SPL values for various microphone positions. The pavement has a limited influence on tire noise emission directivity pattern. The pattern depends mainly on the volume of open and interlinked voids in a pavement but it does not depend on the aggregate size and its kind nor on kind of the binder. There is no difference in noise directivity patterns obtained on three different non-porous pavements, but for porous and poroelastic pavements the pattern depends very much on tire/pavement combination. Porous asphalt concrete (PAC) is the less-directional of all measured pavements. There are no significant differences in noise emission directivity patterns between summer tires 'A' and 'B'. But there are significant differences in directivity patterns of those two and of other tires for several tire/pavement combinations.