This paper focuses on the security analysis of ElephantDelirium algorithm. Elephant is one of the candidate algorithms in the finalist of National Institute of Standards and Technology ( NIST) lightweight cryptographic ( LWC) project. Its encryption algorithm extends the key to the secret masks through an invertible map, and then uses a permutation on the internal states to achieve confusion and diffusion. The ElephantDelirium algorithm is an instance of Elephant encryption algorithm which uses Keccakf[ 200] as its underlying permutation. This paper constructs a 5round zerosum distinguisher using the property that the algebraic degree of nonlinear operation in Keccakf[ 200] permutation is 2. Based on this distinguisher, we use the divide and conquer method to guess the secret mask in the output of 6round ElephantDelirium algorithm and filter out the right secret mask by checking the ze rosum property. As a result, the secret mask can be recovered with 100% accuracy and 100% suc cess rate. This attack is under the noncerespecting setting and costs about 2. 8 seconds to recover all key bits using a single CPU core. This work is the first practical attack on the ElephantDelirium algo rithm. Also, we improve the result of optimized interpolation attack on 8round ElephantDelirium algo rithm with the help of the cube attack. This improvement reduces the complexity from 298. 3 to 295. 2 .