The T1 value of pure water, which is often used as a phantom to simulate cerebrospinal fluid, is significantly different from that of in‐vivo cerebrospinal fluid. The purpose of this study was to develop a phantom with a T1 value equivalent to that of in‐vivo cerebrospinal fluid under examination room temperature (23°C–25°C). In this study, 1.5 and 3.0 T magnetic resonance imaging scanners were used. We examined the signal intensity change in relation to pure water temperature, the T1 values of acetone‐diluted solutions (0–100 v/v%, in 10 steps), and the correlation coefficients obtained from volunteers and the prepared phantoms. The T1 value was close to the value reported in the literature for cerebrospinal fluid when the acetone‐diluted solution was 70 v/v% or higher at scan room temperature. The value at that time was 3532.81–4704.57 ms at 1.5 T and it ranged from 4052.41 to 5701.61 ms at 3.0 T. The highest correlation with the values obtained from the volunteers was r = 0.993 with pure acetone at 1.5 T and r = 0.991 with acetone 90 v/v% at 3.0 T. The relative error of the best phantom‐volunteer match was 32.61 (%) ± 6.71 at 1.5 T and 46.67 (%) ± 4.31 at 3.0 T. The T1 value measured by the null point method did not detect a significant difference between in vivo CSF and acetone 100 v/v% at 1.5 T and acetone 90 v/v% at 3.0 T. The T1 value of cerebrospinal fluid in the living body at scan room temperature was reproduced with acetone. The optimum concentration of acetone for cerebrospinal‐fluid reproduction was pure acetone at 1.5 T and 90 v/v% at 3.0 T.