The creep behaviour of wood or engineered wood products is an important factor in the consideration of the serviceability of a timber structure. This study evaluated the effect of creep of downscaled cross-laminated timber (CLT) specimens by considering panel direction, load level and relative humidity (RH). The test materials were sprucepine-fir (SPF) lumber and polyurethane adhesive. Static bending, one-week creep and four-month creep tests were conducted. It was found that: 1) CLT specimens had an average proportional limit (PL) of around 3,024 N and 817 N in the "Major-strength" and "Minor-strength" directions, respectively; 2) Panel direction strongly impacted time-dependent deflection, with the "Minor-strength" specimens exhibiting a 2 to 2.5 times as large deflection as the "Major-strength" ones; 3) Load level clearly affected the load-deflection behaviour due to the layup of the specimens; 4) Ambient humidity had a little effect on 7-day creep, but impacted the 4-month mechano-sorptive creep to some degree; and 5) The "Generalized Kelvin" model seemed to be suitable in simulation the creep of the downscaled CLT specimens used in this study. The upper limit of the model at 95% confidence level could be a good approach to simulate the creep of the specimens under cyclic RH.