A one-liter stainless steel autoclave equipped with thermometer, magnetic stirrer, gas inlet and gas outlet was charged with 0.125 millimole of Rh4 (CO)12, 50 millimoles of sodium m-(diphenylphosphino)benzenesulfonate, 250 ml of water and 250 ml of sulfolane, and then purged well with a hydrogen-carbon monoxide gaseous mixture (mole ratio 2/1). The contents were heated to 75° C. while maintaining the pressure within the autoclave at 10 atmospheres with the same gaseous mixture and the rate of flow of the effluent gas at 10 liters per hour, and then stirring was continued at a constant temperature of 75° C. for an additional 30 minutes. Thereafter, 64.3 g (500.1 millimoles) of 7-octen-1-al (purity 98%, n-octanal content 2%) was introduced into the autoclave through a metering pump continuously over an hour. After completion of the addition, stirring was continued for 1.5 hours. After the overall 2.5 hours of reaction, stirring was discontinued and the autoclave was cooled. Then, a trace amount of the reaction mixture was sampled and analyzed by gas chromatography. It was found that the amount of 7-octen-1-al was 85.3 millimoles (82.9% conversion) with the yields of 1,9-nonanedial and 2-methyl-1,8-octanedial being 348.5 millimoles and 62.2 millimoles, respectively. The reaction mixture was transferred under pressure to a three-necked 2-liter flask previously purged to a sufficient extent with a hydrogen-carbon monoxide mixture (mole ratio: 1/1) while avoiding contact of said reaction mixture with air. Then 150 ml of n-octanol and 350 ml of hexane were added, and the whole mixture was stirred in an atmosphere of the same gaseous mixture as mentioned above for 20 minutes while maintaining the internal temperature at 30° C. Upon discontinuation of the stirring, the reaction mixture immediately separated into two layers. After allowing the mixture to stand for 10 minutes, the lower (yellow) and upper (colorless) layers were analyzed by gas chromatography and it was found that 95%, 75% and 86% of 7-octen-1-al, 1,9-nonanedial and 2-methyl-1,8-octanedial, respectively, had been transferred to the upper layer. The lower layer (extraction residue) was then transferred to the autoclave while avoiding contact of said layer with air, and 10 ml of sulfolane was added. Under the same conditions as in the first run, 64.3 g of 7-octen-1-al was added continuously over an hour and thereafter stirring was continued for 1.5 hours. Analysis of the reaction mixture revealed that 87.4 millimoles of 7-octen-1-al remained unreacted and 346.8 millimoles of 1,9-nonanedial and 61.9 millimoles of 2-methyl-1,8-octanedial had been formed in the second hydroformylation reaction. The reaction mixture was subjected to the same extraction procedure under the same conditions as in the first run and the extraction residue (catalyst-containing layer) was charged into the autoclave, followed by addition of 10 ml of sulfolane. The third run of hydroformylation was carried out by the same procedure under the same conditions as in the first run. After discontinuation of reaction, the reaction mixture was analyzed and it was found that 92.4 millimoles of 7-octen-1-al remained unreacted and 342.7 millimoles of 1,9-nonanedial and 59.8 millimoles of 2-methyl-1,8-octanedial had been formed in the third hydroformylation reaction. Then, the same extraction procedure as in the first run was performed.