Locomotives of railways worldwide have been using sand since 1838 to improve the wheel–rail adhesion during traction and braking operations. In more recent years, sanders have also been fitted to electrical and diesel multiple units in some railways to fight (often in combination with traction control or wheel slide protection systems) the low-adhesion conditions, especially encountered in autumn due to leaf contamination. In spite of the worldwide broad use of sand, different standards on sanding practice appear to be used by different railways, while there is a lack of fundamental understanding on the influence of sanding parameters, such as particle size distribution, feed rate, and number of sanding axles (among others), on the adhesion recovery, wear, and train detection. In order to gain a better understanding of these interrelationships, the authors have carried out laboratory investigations in recent years. As a continuation of that work, the influence of the sand particle size on the adhesion recovery in leaf-contaminated contacts is investigated in this article by means of traction tests of an electrical locomotive in a stabling yard. Three differently sized silica sands are used in the testing. In addition to the particle size investigations, the standard sand currently used in the Dutch railways is also tested to quantify its effectiveness against leaf contamination. Besides the immediate adhesion improvement upon sanding, the remaining friction level left for subsequent tractive wheel passages is also investigated for all sands tested. Furthermore, baseline tests (i.e. without application of sand) are also performed to obtain some quantitative insight into the impact of leaf contamination on the wheel–rail adhesion in the field.