We examine the duty cycle and the history of star formation ( SFH ) for high-redshift galaxies at z \geq 6 using cosmological hydrodynamic simulations . We find that , even though individual galaxies have bursty SFH , the averaged SFH between z \sim 15 to z = 6 can be characterized well by either an exponentially increasing functional form with characteristic time-scales of 70 Myr to 200 Myr for galaxies with stellar masses M _ { s } \sim 10 ^ { 6 } M _ { \odot } to > 10 ^ { 10 } M _ { \odot } respectively , or by a simple power-law form which exhibits a similar mass dependent time-scales . Using the SFH of individual galaxies , we measure the duty cycle of star formation ( DC _ { SFH } ) ; i.e. , the fraction of time a galaxy of a particular mass spends above a star formation rate ( SFR ) threshold which would make it observable to the Hubble Space Telescope ( HST ) during a given epoch . We also examine the fraction of galaxies at a given redshift that are brighter than a rest-frame UV magnitude ( M _ { uv } \sim - 18 ) , which is sufficient enough to make them observable ( DC _ { M _ { uv } } ) . We find that both DC _ { SFH } and DC _ { M _ { uv } } make a sharp transition from zero ( for galaxies with M _ { s } \lesssim 10 ^ { 7 } M _ { \odot } ) to unity ( for M _ { s } > 10 ^ { 9 } M _ { \odot } ) . The measured duty cycle is also manifested in the intrinsic scatter in the M _ { s } - SFR relationship ( \sim 1 dex ) and M _ { s } - M _ { uv } relationship ( \Delta M _ { uv } \sim \pm 1 mag ) . We provide analytic fits to the DC as a function of M _ { s } using a sigmoid function which can be used to correct for catalogue incompleteness . We consider the effects of duty cycle to the observational estimate of galaxy stellar mass functions ( GSMF ) and the star formation rate density ( SFRD ) , and find that it results in a much shallower low-mass end slopes of the GSMF and a reduction of \gtrsim 70 % of our intrinsic SFRD , making our simulation results more compatible with observational estimates .