Cerebrospinal fluid (CSF) cytology is an effective tool for evaluating diseases involving the central nervous system, but this technique is usually limited by its low cellularity and poor cellular preservation. Here we compared the manual liquid-base $Liqui-PREP^{TM}$ (LP) to the cytospin (CS) with using a mononuclear cell suspension and we applied both methods to the CSFs of pediatric leukemia patients. The cytopresevability, in terms of cell yield and cell size, and the clinical efficacy were evaluated. When 2000 and 4000 mononuclear cells were applied, LP was superior to CS for the cell yield, 16.8% vs 1.7% (P=0.001) and 26.2% vs 3.5% (P=0.002), respectively. The mean size of the smeared cells was 10.60 ${mu}m$ in the CS, 5.01 ${mu}m$ in the LP and 6.50 ${mu}m$ in the direct smear (DS), and the size ratio was 1.7 (CS to DS), 0.8(LP to DS) and 2.1 (CS to LP), respectively. As compared to the cells in the DS, the cells in the CS were significantly enlarged, but those in the LP were slightly shrunken. Upon application to 109 CSF samples, 4 were diagnosed as positive for leukemia (positive), 4 had atypical cells and 101 were negative by CS; 6 were positive, one had atypical cells and 102 were negative by LP. For six cases, in which 4 were positive for leukemia and 2 of 4 had atypical cells by CS, they were positive by LP and they were also confirmed as positive according to the follow-up study. Three cases diagnosed as atypical cells (two by CS and one by LP), were confirmed as negative. In conclusion, these results suggest that LP is superior to CS for the cytopresevability and for rendering a definite diagnosis of cerebrospinal fluid.