One of the unknown parameters in neutrino oscillations is the octant of the mixing angle theta_{23}. In this paper, we discuss the possibility of determining the octant of theta_{23} in the long baseline experiments T2K and NOvA in conjunction with future atmospheric neutrino detectors, in light of non-zero value of theta_{13} measured by reactor experiments. We consider two detector technologies for atmospheric neutrinos - magnetized iron calorimeter and non-magnetized Liquid Argon Time Projection Chamber. We present the octant sensitivity for T2K/NOvA and atmospheric neutrino experiments separately as well as combined. For the long baseline experiments, a precise measurement of theta_{13}, which can exclude degenerate solutions in the wrong octant, increases the sensitivity drastically. For theta_{23} = 39^o and sin^2 2 theta_{13} = 0.1, at least ~2 sigma sensitivity can be achieved by T2K+NOvA for all values of delta_{CP} for both normal and inverted hierarchy. For atmospheric neutrinos, the moderately large value of theta_{13} measured in the reactor experiments is conducive to octant sensitivity because of enhanced matter effects. A magnetized iron detector can give a 2 sigma octant sensitivity for 500 kT yr exposure for theta_{23} = 39^o, delta_{CP} = 0 and normal hierarchy. This increases to 3 sigma for both hierarchies by combining with T2K+NOvA. This is due to a preference of different theta_{23} values at the minimum chi^2 by T2K/NOvA and atmospheric neutrino experiments. A Liquid Argon detector for atmospheric neutrinos with the same exposure can give higher octant sensitivity, due to the interplay of muon and electron contributions and superior resolutions. We obtain a ~3 sigma sensitivity for theta_{23} = 39^o for normal hierarchy. This increases to > ~4 sigma for all values of delta_{CP} if combined with T2K+NOvA. For inverted hierarchy the combined sensitivity is ~3 sigma.