This study was initiated because of curiosity as to the reasons for the occasional development of severe arachnoiditis following myelography with ethyl iodophenylundecylate (Pantopaque; Myodil). There are numerous reports in the literature of this occurrence with operative (2, 7–10, 12) or autopsy (1, 3, 5) confirmation of severe meningeal inflammation. The present communication represents a continuing study of experimentally produced arachnoiditis in dogs which was first reported in March 1963 (4). Our initial convictions relating to the hazards of arachnoiditis due to the intrathecal injection of Pantopaque in the presence of subarachnoid bleeding were stimulated by Jaeger's studies (6). He suggested that blood serum, acting as an emulsifying agent when mixed with iodized oil, might disperse the oil as minute irritating droplets. Because of difficulties in ascertaining whether or not subarachnoid bleeding had occurred in human clinical material, it was decided to evaluate the effect of blood mixed with Pantopaque in the subarachnoid space of dogs. Initial studies of this problem (4) prompted us to expand these investigations. Following the publication by Sehgal et al. of successful treatment of chronic symptomatic Pantopaque arachnoiditis with subarachnoid injections of corticosteroids (11), we decided to investigate the possibility of prevention of arachnoiditis in experimental animals. Materials and Methods Twenty-four mongrel dogs in good clinical condition were used. In all, the material was injected into the cisterna magna through a posterior midline approach with a 20-gauge infant spinal needle, with strict aseptic technic. Injection was made with the animal in a 60° caudal tilt and the neck flexed. The neck was extended immediately upon completion of the injection. In this position, almost all the material flowed slowly along the subarachnoid space to the lumbar region over a period of approximately one hour. The injection was made slowly after preliminary withdrawal of 5 ml of spinal fluid. Radiographs were obtained at intervals until the medium reached the lumbar region. The animal was maintained under light pentobarbital anesthesia during this period. Two dogs were injected with 3 ml of blood alone, 4 with 3 ml of Pantopaque, 6 with 3 ml of Pantopaque and 3 ml of blood. Six more animals were injected with 3 ml of Pantopaque and 20 mg of methylprednisolone acetate, and 6 with 3 ml of Pantopaque plus 3 ml of blood plus 20 mg of methylprednisolone acetate. The dogs were sacrificed six to eight weeks following injection, with the exception of 2 animals which had to be sacrificed earlier because of extremely poor clinical status. Radiographs obtained at the time of sacrifice were compared with the initial films to estimate the amount of Pantopaque absorbed from the subarachnoid space in this interval.