We report doping of green fluorescent protein from an electrospray ionization (ESI) source into superfluid helium droplets. From analyses of the time profiles of the doped droplets, we identify two distinct groups of droplets. The faster group has a smaller average size, on the order of 10(6) helium atoms/droplet, and the slower group is much larger, by at least an order of magnitude. The relative populations of these two groups depend on the temperature of the droplet source: from 11 to 5 K, the signal intensity of the slower droplet group gradually increases, from near the detection limit to comparable to that of the faster group. We postulate that the smaller droplets are formed via condensation of gaseous helium upon expansion from the pulsed valve, while the larger droplets develop from fragmentation of ejected liquid helium. Our results on the size and velocity of the condensation peak at higher source temperatures (>7 K) agree with previous reports, but those at lower temperatures (<7 K) seem to be off. We attribute this discrepancy to the masking effect of the exceedingly large droplets from the fragmentation peak in previous measurements of droplet sizes. Within the temperature range of our investigation, although the expansion condition changes from subcritical to supercritical, there is no abrupt change in either the velocity distribution or the size distribution of the condensation peak, and the most salient effect is in the increasing intensity of the fragmentation peak. The absolute doping efficiency, as expressed by the ratio of ion-doped droplets over the total number of ions from the ESI source, is on the order of 10(-4), while only hundreds of doped ions have been detected. Further improvements in the ESI source are key to extending the technology for future experiments. On the other hand, the separation of the two groups of droplets in velocity is beneficial for size selection of only the smaller droplets for future experiments of electron diffraction.