Context : The ultra-compact dipping source XB 1916-053 has an orbital period of close to 50 min and a companion star with a very low mass ( less than 0.1 M _ { \odot } ) .
The orbital period derivative of the source was estimated to be 1.5 ( 3 ) \times 10 ^ { -11 } s/s through analysing the delays associated with the dip arrival times obtained from observations spanning 25 years , from 1978 to 2002 .

Aims : The known orbital period derivative is extremely large and can be explained by invoking an extreme , non-conservative mass transfer rate that is not easily justifiable .
We extended the analysed data from 1978 to 2014 , by spanning 37 years , to verify whether a larger sample of data can be fitted with a quadratic term or a different scenario has to be considered .

Methods : We obtained 27 delays associated with the dip arrival times from data covering 37 years and used different models to fit the time delays with respect to a constant period model .

Results : We find that the quadratic form alone does not fit the data .
The data are well fitted using a sinusoidal term plus a quadratic function or , alternatively , with a series of sinusoidal terms that can be associated with a modulation of the dip arrival times due to the presence of a third body that has an elliptical orbit .
We infer that for a conservative mass transfer scenario the modulation of the delays can be explained by invoking the presence of a third body with mass between 0.10–0.14 M _ { \odot } , orbital period around the X-ray binary system of close to 51 yr and an eccentricity of 0.28 \pm 0.15 .
In a non-conservative mass transfer scenario we estimate that the fraction of matter yielded by the degenerate companion star and accreted onto the neutron star is \beta = 0.08 , the neutron star mass is \geq 2.2 M _ { \odot } , and the companion star mass is 0.028 M _ { \odot } .
In this case , we explain the sinusoidal modulation of the delays by invoking the presence of a third body with orbital period of 26 yr and mass of 0.055 M _ { \odot } .

Conclusions : From the analysis of the delays associated with the dip arrival times , we find that both in a conservative and non-conservative mass transfer scenario we have to invoke the presence of a third body to explain the observed sinusoidal modulation .
We propose that XB 1916-053 forms a hierarchical triple system .