Objective   To evaluate the reproducibility and agreement in the measurement of liver volume with multi-slice spiral computed tomography (CT).     Methods   Spiral CT was performed in 20 cases (14 healthy volunteers and 6 patients with cirrhosis). Images were transferred to a workstation

and volumes were measured by three observers with 5-year abdominal CT experience. Each observer perform

ed the measurements in all subjects twice. Intra-observer and inter-observer variability was calculated using analysis of variance of the repeatedly measured results. The agreement was analyzed with the intra-class correlation coefficient (ICC).  Reliability analysis in inter-observer was performed with Cronbach’s alpha coefficient.    Results   Data measured by Observer A were (1 265.53±421.43)cm

3

and (1 272.30± 414.73)cm

3

[difference: (0.80±2.20)%

P=0.15

ICC=0.998 8

]

Observer B (1 299.97±403.48)cm

3

and (1 310.23±404.04)cm

3

[difference: (0.90±1.90)%

   P=0.07

ICC=0.998 3

]

Observer C (1 282.18±425.91)cm

3

and (1 289.73±424.66)cm

3

[difference: (0.70±2.60)%

  P=0.19

ICC=0.998 3

]

respectively. The mean liver volumes by the three observers were (1 268.91±418.07)cm

3

(1 305.10±403.59)cm

3

and (1 285.97±425.10)cm

3

respectively (P=0.07

ICC=0.986 9

Cronbach’s alpha coefficient=0.995 6).    Conclusion    High reproducibility and agreement in the measurement of liver volume can be achieved with multi-slice spiral CT.