Fluorescence in situ hybridization (FISH) can identify chromosomal translocations on fixed archival tissue, but studies cross-validating the utility of FISH on lesions of different cell lineages that harbor similar translocations (e.g. those involving anaplastic lymphoma kinase [ALK]) have not been published. Our objective was to define the diagnostic utility, performance characteristics, and limitations of a commercially available, split-signal, FISH probe for ALK gene rearrangements on fixed, archived tissue from lesions of diverse cell lineage. The sensitivity, specificity, and positive and negative predictive values of the Vysis ALK FISH probe were compared with those of the ALK-1 antibody (Dako) in a series of 101 cases, comprising 43 hematolymphoid neoplasms, 4 reactive lymphoid controls, 50 non-hematolymphoid (including neuroectodermal, epithelial, myofibroblastic, and germ cell) lesions, and 4 early-trimester aborted fetuses that served as neuroblastic controls. The study involved a predominantly (72%) Singaporean Chinese population aged between 9 months and 88 years (excluding the aborted fetal controls). All cases were reviewed both histologically and immunohistochemically with a wide panel of antibodies using the standard protocols in order to diagnose them according to the latest WHO classification systems. A positive cut-off value was determined, both by comparison with diagnostic categories with and without ALK translocations, as well as with negative controls. The ALK FISH probe suffered a 33% non-informative rate, but in informative cases it showed 94% concordance with the ALK-1 immunostain. A minimum cut-off value of 5 in 200 informative cells was adopted to make a positive call in each case. Of the ALK-1 immunoreactive lesions, nine lymphomas were concordantly ALK translocation-positive but one vesical inflammatory myofibroblastic tumor was discordantly FISH-negative. Among the ALK-1-immunonegative lesions, one case each of anaplastic lymphoma and pulmonary mycobacterial spindle cell pseudotumor were discordantly ALK FISH-positive, while a case each of intestinal myeloblastic tumor and ganglioglioma showed initial--but not reproducible--positive FISH readings. The remaining cases were concordantly negative. The discrepancies between ALK FISH results and well-established immunomorphological parameters indicate that interpretation is not always straightforward. Notably, the derivation of threshold cut-off values for positive calls on FISH assays has seldom been addressed in the literature, and has raised issues in interpreting cases with borderline positivity in this study. The factors that may influence such cut-off values are extensively reviewed. We propose the term 'conditional threshold positivity' to encourage the adoption of different cut-off values for making positive calls in lesions of different origin.