The only way to detect planets around stars at distances of several kpc is by (photometric or astrometric) microlensing observations. In this paper, we show that the capability of photometric microlensing extends to the detection of signals caused by planets around stars in nearby galaxies (e.g. M31) and that there is no other method that can achieve this. Due to the large crowding, microlensing experiments towards M31 can only observe the high-magnification part of a lensing light curve. Therefore, the dominating channel for microlensing signals by planets is in distortions near the peak of high-magnification events as discussed by Griest and Safizadeh. We calculate the probability to detect planetary anomalies for microlensing experiments towards M31 and find that jupiter-like planets around stars in M31 can be detected. Though the characterization of the planet(s) involved in this signal will be difficult, the absence of such signals can yield strong constraints on the abundance of jupiter-like planets.