We present results from an analysis of all data taken by the BICEP2 / Keck CMB polarization experiments up to and including the 2015 observing season .
This includes the first Keck Array observations at 220 GHz and additional observations at 95 & 150 GHz .
The Q / U maps reach depths of 5.2 , 2.9 and 26 \mu { \mathrm { K } } _ { \mathrm { \mbox { \tiny \sc cmb } } } arcmin at 95 , 150 and 220 GHz respectively over an effective area of \approx 400 square degrees .
The 220 GHz maps achieve a signal-to-noise on polarized dust emission approximately equal to that of Planck at 353 GHz .
We take auto- and cross-spectra between these maps and publicly available WMAP and Planck maps at frequencies from 23 to 353 GHz .
We evaluate the joint likelihood of the spectra versus a multicomponent model of lensed- \Lambda CDM+ r +dust+synchrotron+noise .
The foreground model has seven parameters , and we impose priors on some of these using external information from Planck and WMAP derived from larger regions of sky .
The model is shown to be an adequate description of the data at the current noise levels .
The likelihood analysis yields the constraint r _ { 0.05 } < 0.07 at 95 % confidence , which tightens to r _ { 0.05 } < 0.06 in conjunction with Planck temperature measurements and other data .
The lensing signal is detected at 8.8 \sigma significance .
Running maximum likelihood search on simulations we obtain unbiased results and find that \sigma ( r ) = 0.020 .
These are the strongest constraints to date on primordial gravitational waves .