A multi-spectral detector was developed to evaluate crop growth condition. The detector mainly consists of three parts, a fiber, two sensors, and a MCU. Since two wavelengths were usually used to form an index to evaluate crop growth condition, the fiber was designed to be in "Y" type in order to measure crop reflectance at two different wavelengths. The fiber collect reflect light in one side from crop leaf or canopy and the collected light was divided into two parts and introduced to two sensors. The sensor is composed of a light guiding pipe, an inlet, two gaskets, an interferential filter, a photoelectric cell, and others. It can form a closed space in the center in order to measure reflecting light without external noise. Two different filters were used in two sensors with different special transmission wavelengths so that the reflect lights at wanted wavelengths could be obtained. The MCU was used to amplify signal, display measurement, and record data. It can calculate an evaluation index of crop growth condition from measured reflectance by an embedded evaluation model. It can also transmit data to PC or other storage device. Performance test was conducted. The wavelengths of two interferential filters were selected as 527 nm and 762 nm based on the result of field test. The result shows that the detector can measure the spectral reflectance of crop leaf in high sensitivity and accuracy. And the Normalized Difference Color Index calculated from the reflectance can be used to evaluate crop growth condition well.