Within the last years pollen grains have gained increasing attention due to their cloud-forming potential. Especially the discovery that ice nucleating macromolecules (INMs) or subpollen particles (SPPs) obtained from pollen grains are able to initiate freezing has stirred up interest in pollen. INMs and SPPs are much smaller and potentially more numerous than pollen grains and could significantly affect cloud formation in the atmosphere. However, INMs and SPPs are not clearly distinguished. This has motivated the present study, which focuses on birch pollen and investigates the relationship between pollen grains, INMs, and SPPs. According to the usage of the term SPP in the medical fields, we define SPPs as the starch granules contained in pollen grains. We show that these insoluble SPPs are only obtained when fresh pollen grains are used to generate aqueous extracts from pollen. Due to the limited seasonal availability of fresh pollen grains, almost all studies have been conducted with commercial pollen grains. To enable the investigation of the SPPs we develop an alternative extraction method to generate large quantities of SPPs from commercial pollen grains. We show that INMs are not bonded to SPPs (i.e. can be washed off with water). Further, we find that purified SPPs are not ice nucleation active: after several times of washing SPPs with ultrapure water the ice nucleation activity completely disappears. To our knowledge, this is the first study to investigate the ice nucleation activity of isolated SPPs. To study the chemical nature of the INMs, we use fluorescence spectroscopy. Fluorescence excitation–emission maps indicate a strong signal in the protein range (maximum around λex = 280 nm and λem = 330 nm) with all ice nucleation active samples. In contrast, with purified SPPs the protein signal is lost. We also quantify the protein concentration with the Bradford assay. The protein concentration ranges from 77.4 µg mL−1 (highly concentrated INMs) to below 2.5 µg mL−1 (purified SPPs). Moreover, we investigate the connection between proteins and ice nucleation activity by treating the ice nucleation active samples with subtilisin A and urea to unfold and digest the proteins. After this treatment the ice nucleation activity clearly diminished. The results indicate a linkage between ice nucleation activity and protein concentration. The missing piece of the puzzle could be a glycoprotein which exhibits carboxylate functionalities, can bind water in tertiary structures, and displays degeneration and unfolding of its secondary structure due to heat treatment or reaction with enzymes. Even though purified SPPs are not ice nucleation active they could act as carriers of INMs and distribute those in the atmosphere.