It is well known that clinical hypothyroidism (CH) can induce cognitive deficits, and the decision to start treatment for CH with thyroxine is usually straightforward. However, the relationship of cognition dysfunction with subclinical hypothyroidism (SCH) is inconsistent, and the decision concerning the need to treat SCH is controversial. In the present study, we induced a SCH rat model by hemi-thyroid electrocauterisation; then employed a serial of behavioural tests, including a beam balance, open field task and Morris water maze (MWM), to investigate the behaviour performance of SCH rats; and finally explored the protein expression of phosphorylated extracellular signal-regulated kinase (ERK)1/2 in the hippocampus by western blotting. The results demonstrated that hemi-thyroid electrocauterised rats had an elevated plasma thyrotrophin-stimulating hormone (TSH) level, with normal free thyroxine (fT4) and triiodothyronine (T3) concentrations, which defines SCH in humans. If rat SCH is diagnosed according to measurements of both plasma TSH higher than 97.5 percentile for the sham group and fT4 in the range 2.5–97.5 percentile for the sham group, the success rate of SCH modelling was 66.6%. SCH decreased exploratory behaviour but did not affect motor function in rats, showing a negative correlation of exploratory behaviour with plasma TSH concentration. Moreover, SCH rats displayed an impairment of learning and memory ability in the MWM task, with a longer escape latency in the acquisition phase and a shorter duration in the target quadrant in the test phase compared to that of sham rats. The mechanism for this might be related to the increased plasma TSH concentration, the decreased hippocampal T3 level and the enhanced expression of phosphorylated ERK1/2 in the hippocampus. The results of the present study, together with the results obtained in other studies, suggest that treatment is necessary for SCH.